Commit | Line | Data |
a0d0e21e |
1 | /* regexec.c |
2 | */ |
3 | |
4 | /* |
5 | * "One Ring to rule them all, One Ring to find them..." |
6 | */ |
7 | |
a687059c |
8 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
9 | * confused with the original package (see point 3 below). Thanks, Henry! |
10 | */ |
11 | |
12 | /* Additional note: this code is very heavily munged from Henry's version |
13 | * in places. In some spots I've traded clarity for efficiency, so don't |
14 | * blame Henry for some of the lack of readability. |
15 | */ |
16 | |
e50aee73 |
17 | /* The names of the functions have been changed from regcomp and |
18 | * regexec to pregcomp and pregexec in order to avoid conflicts |
19 | * with the POSIX routines of the same names. |
20 | */ |
21 | |
b9d5759e |
22 | #ifdef PERL_EXT_RE_BUILD |
23 | /* need to replace pregcomp et al, so enable that */ |
24 | # ifndef PERL_IN_XSUB_RE |
25 | # define PERL_IN_XSUB_RE |
26 | # endif |
27 | /* need access to debugger hooks */ |
28 | # ifndef DEBUGGING |
29 | # define DEBUGGING |
30 | # endif |
31 | #endif |
32 | |
33 | #ifdef PERL_IN_XSUB_RE |
d06ea78c |
34 | /* We *really* need to overwrite these symbols: */ |
56953603 |
35 | # define Perl_regexec_flags my_regexec |
36 | # define Perl_regdump my_regdump |
37 | # define Perl_regprop my_regprop |
d06ea78c |
38 | /* *These* symbols are masked to allow static link. */ |
39 | # define Perl_pregexec my_pregexec |
d88dccdf |
40 | # define Perl_reginitcolors my_reginitcolors |
56953603 |
41 | #endif |
42 | |
f0fcb552 |
43 | /*SUPPRESS 112*/ |
a687059c |
44 | /* |
e50aee73 |
45 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
46 | * |
47 | * Copyright (c) 1986 by University of Toronto. |
48 | * Written by Henry Spencer. Not derived from licensed software. |
49 | * |
50 | * Permission is granted to anyone to use this software for any |
51 | * purpose on any computer system, and to redistribute it freely, |
52 | * subject to the following restrictions: |
53 | * |
54 | * 1. The author is not responsible for the consequences of use of |
55 | * this software, no matter how awful, even if they arise |
56 | * from defects in it. |
57 | * |
58 | * 2. The origin of this software must not be misrepresented, either |
59 | * by explicit claim or by omission. |
60 | * |
61 | * 3. Altered versions must be plainly marked as such, and must not |
62 | * be misrepresented as being the original software. |
63 | * |
64 | **** Alterations to Henry's code are... |
65 | **** |
4eb8286e |
66 | **** Copyright (c) 1991-1999, Larry Wall |
a687059c |
67 | **** |
9ef589d8 |
68 | **** You may distribute under the terms of either the GNU General Public |
69 | **** License or the Artistic License, as specified in the README file. |
a687059c |
70 | * |
71 | * Beware that some of this code is subtly aware of the way operator |
72 | * precedence is structured in regular expressions. Serious changes in |
73 | * regular-expression syntax might require a total rethink. |
74 | */ |
75 | #include "EXTERN.h" |
864dbfa3 |
76 | #define PERL_IN_REGEXEC_C |
a687059c |
77 | #include "perl.h" |
0f5d15d6 |
78 | |
a687059c |
79 | #include "regcomp.h" |
80 | |
c277df42 |
81 | #define RF_tainted 1 /* tainted information used? */ |
82 | #define RF_warned 2 /* warned about big count? */ |
ce862d02 |
83 | #define RF_evaled 4 /* Did an EVAL with setting? */ |
a0ed51b3 |
84 | #define RF_utf8 8 /* String contains multibyte chars? */ |
85 | |
86 | #define UTF (PL_reg_flags & RF_utf8) |
ce862d02 |
87 | |
88 | #define RS_init 1 /* eval environment created */ |
89 | #define RS_set 2 /* replsv value is set */ |
c277df42 |
90 | |
a687059c |
91 | #ifndef STATIC |
92 | #define STATIC static |
93 | #endif |
94 | |
c277df42 |
95 | /* |
96 | * Forwards. |
97 | */ |
98 | |
ae5c130c |
99 | #define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c)) |
a0ed51b3 |
100 | #define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p)) |
101 | |
102 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
103 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) |
104 | |
dfe13c55 |
105 | #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off)) |
106 | #define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off)) |
107 | #define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off)) |
108 | #define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off)) |
109 | #define HOPc(pos,off) ((char*)HOP(pos,off)) |
110 | #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) |
a0d0e21e |
111 | |
76e3520e |
112 | STATIC CHECKPOINT |
cea2e8a9 |
113 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e |
114 | { |
11343788 |
115 | dTHR; |
3280af22 |
116 | int retval = PL_savestack_ix; |
117 | int i = (PL_regsize - parenfloor) * 4; |
a0d0e21e |
118 | int p; |
119 | |
120 | SSCHECK(i + 5); |
3280af22 |
121 | for (p = PL_regsize; p > parenfloor; p--) { |
cf93c79d |
122 | SSPUSHINT(PL_regendp[p]); |
123 | SSPUSHINT(PL_regstartp[p]); |
3280af22 |
124 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e |
125 | SSPUSHINT(p); |
126 | } |
3280af22 |
127 | SSPUSHINT(PL_regsize); |
128 | SSPUSHINT(*PL_reglastparen); |
129 | SSPUSHPTR(PL_reginput); |
a0d0e21e |
130 | SSPUSHINT(i + 3); |
131 | SSPUSHINT(SAVEt_REGCONTEXT); |
132 | return retval; |
133 | } |
134 | |
c277df42 |
135 | /* These are needed since we do not localize EVAL nodes: */ |
c3464db5 |
136 | # define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, \ |
137 | " Setting an EVAL scope, savestack=%i\n", \ |
3280af22 |
138 | PL_savestack_ix)); lastcp = PL_savestack_ix |
c3464db5 |
139 | |
3280af22 |
140 | # define REGCP_UNWIND DEBUG_r(lastcp != PL_savestack_ix ? \ |
c3464db5 |
141 | PerlIO_printf(Perl_debug_log, \ |
142 | " Clearing an EVAL scope, savestack=%i..%i\n", \ |
3280af22 |
143 | lastcp, PL_savestack_ix) : 0); regcpblow(lastcp) |
c277df42 |
144 | |
76e3520e |
145 | STATIC char * |
cea2e8a9 |
146 | S_regcppop(pTHX) |
a0d0e21e |
147 | { |
11343788 |
148 | dTHR; |
a0d0e21e |
149 | I32 i = SSPOPINT; |
150 | U32 paren = 0; |
151 | char *input; |
cf93c79d |
152 | I32 tmps; |
a0d0e21e |
153 | assert(i == SAVEt_REGCONTEXT); |
154 | i = SSPOPINT; |
155 | input = (char *) SSPOPPTR; |
3280af22 |
156 | *PL_reglastparen = SSPOPINT; |
157 | PL_regsize = SSPOPINT; |
c277df42 |
158 | for (i -= 3; i > 0; i -= 4) { |
a0d0e21e |
159 | paren = (U32)SSPOPINT; |
3280af22 |
160 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d |
161 | PL_regstartp[paren] = SSPOPINT; |
162 | tmps = SSPOPINT; |
3280af22 |
163 | if (paren <= *PL_reglastparen) |
164 | PL_regendp[paren] = tmps; |
c277df42 |
165 | DEBUG_r( |
c3464db5 |
166 | PerlIO_printf(Perl_debug_log, |
167 | " restoring \\%d to %d(%d)..%d%s\n", |
cf93c79d |
168 | paren, PL_regstartp[paren], |
169 | PL_reg_start_tmp[paren] - PL_bostr, |
170 | PL_regendp[paren], |
3280af22 |
171 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 |
172 | ); |
a0d0e21e |
173 | } |
c277df42 |
174 | DEBUG_r( |
3280af22 |
175 | if (*PL_reglastparen + 1 <= PL_regnpar) { |
c3464db5 |
176 | PerlIO_printf(Perl_debug_log, |
177 | " restoring \\%d..\\%d to undef\n", |
3280af22 |
178 | *PL_reglastparen + 1, PL_regnpar); |
c277df42 |
179 | } |
180 | ); |
3280af22 |
181 | for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) { |
182 | if (paren > PL_regsize) |
cf93c79d |
183 | PL_regstartp[paren] = -1; |
184 | PL_regendp[paren] = -1; |
a0d0e21e |
185 | } |
186 | return input; |
187 | } |
188 | |
0f5d15d6 |
189 | STATIC char * |
cea2e8a9 |
190 | S_regcp_set_to(pTHX_ I32 ss) |
0f5d15d6 |
191 | { |
46124e9e |
192 | dTHR; |
0f5d15d6 |
193 | I32 tmp = PL_savestack_ix; |
194 | |
195 | PL_savestack_ix = ss; |
196 | regcppop(); |
197 | PL_savestack_ix = tmp; |
942e002e |
198 | return Nullch; |
0f5d15d6 |
199 | } |
200 | |
201 | typedef struct re_cc_state |
202 | { |
203 | I32 ss; |
204 | regnode *node; |
205 | struct re_cc_state *prev; |
206 | CURCUR *cc; |
207 | regexp *re; |
208 | } re_cc_state; |
209 | |
c277df42 |
210 | #define regcpblow(cp) LEAVE_SCOPE(cp) |
a0d0e21e |
211 | |
a687059c |
212 | /* |
e50aee73 |
213 | * pregexec and friends |
a687059c |
214 | */ |
215 | |
216 | /* |
c277df42 |
217 | - pregexec - match a regexp against a string |
a687059c |
218 | */ |
c277df42 |
219 | I32 |
864dbfa3 |
220 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 |
221 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 |
222 | /* strend: pointer to null at end of string */ |
223 | /* strbeg: real beginning of string */ |
224 | /* minend: end of match must be >=minend after stringarg. */ |
225 | /* nosave: For optimizations. */ |
226 | { |
227 | return |
228 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
229 | nosave ? 0 : REXEC_COPY_STR); |
230 | } |
0f5d15d6 |
231 | |
232 | STATIC void |
cea2e8a9 |
233 | S_cache_re(pTHX_ regexp *prog) |
0f5d15d6 |
234 | { |
46124e9e |
235 | dTHR; |
0f5d15d6 |
236 | PL_regprecomp = prog->precomp; /* Needed for FAIL. */ |
237 | #ifdef DEBUGGING |
238 | PL_regprogram = prog->program; |
239 | #endif |
240 | PL_regnpar = prog->nparens; |
241 | PL_regdata = prog->data; |
242 | PL_reg_re = prog; |
243 | } |
22e551b9 |
244 | |
9661b544 |
245 | STATIC void |
cea2e8a9 |
246 | S_restore_pos(pTHX_ void *arg) |
35ef4773 |
247 | { |
248 | dTHR; |
cf93c79d |
249 | if (PL_reg_eval_set) { |
250 | if (PL_reg_oldsaved) { |
251 | PL_reg_re->subbeg = PL_reg_oldsaved; |
252 | PL_reg_re->sublen = PL_reg_oldsavedlen; |
253 | RX_MATCH_COPIED_on(PL_reg_re); |
254 | } |
9661b544 |
255 | PL_reg_magic->mg_len = PL_reg_oldpos; |
256 | PL_reg_eval_set = 0; |
5c5e4c24 |
257 | PL_curpm = PL_reg_oldcurpm; |
9661b544 |
258 | } |
259 | } |
260 | |
261 | |
a687059c |
262 | /* |
c277df42 |
263 | - regexec_flags - match a regexp against a string |
a687059c |
264 | */ |
79072805 |
265 | I32 |
864dbfa3 |
266 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
22e551b9 |
267 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) |
c277df42 |
268 | /* strend: pointer to null at end of string */ |
269 | /* strbeg: real beginning of string */ |
270 | /* minend: end of match must be >=minend after stringarg. */ |
271 | /* data: May be used for some additional optimizations. */ |
272 | /* nosave: For optimizations. */ |
a687059c |
273 | { |
5c0ca799 |
274 | dTHR; |
a0d0e21e |
275 | register char *s; |
c277df42 |
276 | register regnode *c; |
a0d0e21e |
277 | register char *startpos = stringarg; |
278 | register I32 tmp; |
c277df42 |
279 | I32 minlen; /* must match at least this many chars */ |
a0d0e21e |
280 | I32 dontbother = 0; /* how many characters not to try at end */ |
281 | CURCUR cc; |
c277df42 |
282 | I32 start_shift = 0; /* Offset of the start to find |
a0ed51b3 |
283 | constant substr. */ /* CC */ |
284 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
c277df42 |
285 | I32 scream_pos = -1; /* Internal iterator of scream. */ |
286 | char *scream_olds; |
3280af22 |
287 | SV* oreplsv = GvSV(PL_replgv); |
a687059c |
288 | |
a0d0e21e |
289 | cc.cur = 0; |
4633a7c4 |
290 | cc.oldcc = 0; |
3280af22 |
291 | PL_regcc = &cc; |
a0d0e21e |
292 | |
0f5d15d6 |
293 | cache_re(prog); |
a0d0e21e |
294 | #ifdef DEBUGGING |
3280af22 |
295 | PL_regnarrate = PL_debug & 512; |
a0d0e21e |
296 | #endif |
297 | |
298 | /* Be paranoid... */ |
299 | if (prog == NULL || startpos == NULL) { |
cea2e8a9 |
300 | Perl_croak(aTHX_ "NULL regexp parameter"); |
a0d0e21e |
301 | return 0; |
302 | } |
303 | |
c277df42 |
304 | minlen = prog->minlen; |
305 | if (strend - startpos < minlen) goto phooey; |
306 | |
a0d0e21e |
307 | if (startpos == strbeg) /* is ^ valid at stringarg? */ |
3280af22 |
308 | PL_regprev = '\n'; |
a0d0e21e |
309 | else { |
a0ed51b3 |
310 | PL_regprev = (U32)stringarg[-1]; |
3280af22 |
311 | if (!PL_multiline && PL_regprev == '\n') |
312 | PL_regprev = '\0'; /* force ^ to NOT match */ |
a0d0e21e |
313 | } |
bbce6d69 |
314 | |
a0d0e21e |
315 | /* Check validity of program. */ |
22c35a8c |
316 | if (UCHARAT(prog->program) != REG_MAGIC) { |
cea2e8a9 |
317 | Perl_croak(aTHX_ "corrupted regexp program"); |
a0d0e21e |
318 | } |
319 | |
3280af22 |
320 | PL_reg_flags = 0; |
321 | PL_reg_eval_set = 0; |
a0d0e21e |
322 | |
a0ed51b3 |
323 | if (prog->reganch & ROPT_UTF8) |
324 | PL_reg_flags |= RF_utf8; |
325 | |
326 | /* Mark beginning of line for ^ and lookbehind. */ |
327 | PL_regbol = startpos; |
328 | PL_bostr = strbeg; |
9661b544 |
329 | PL_reg_sv = sv; |
a0ed51b3 |
330 | |
331 | /* Mark end of line for $ (and such) */ |
332 | PL_regeol = strend; |
333 | |
334 | /* see how far we have to get to not match where we matched before */ |
335 | PL_regtill = startpos+minend; |
336 | |
0f5d15d6 |
337 | /* We start without call_cc context. */ |
338 | PL_reg_call_cc = 0; |
339 | |
a0d0e21e |
340 | /* If there is a "must appear" string, look for it. */ |
341 | s = startpos; |
c277df42 |
342 | if (!(flags & REXEC_CHECKED) |
343 | && prog->check_substr != Nullsv && |
774d564b |
344 | !(prog->reganch & ROPT_ANCH_GPOS) && |
c277df42 |
345 | (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL)) |
3280af22 |
346 | || (PL_multiline && prog->check_substr == prog->anchored_substr)) ) |
a0d0e21e |
347 | { |
a0ed51b3 |
348 | char *t; |
349 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
c277df42 |
350 | /* Should be nonnegative! */ |
cf93c79d |
351 | end_shift = minlen - start_shift - |
352 | CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0); |
22e551b9 |
353 | if (flags & REXEC_SCREAM) { |
cf93c79d |
354 | SV *c = prog->check_substr; |
355 | |
356 | if (PL_screamfirst[BmRARE(c)] >= 0 |
357 | || ( BmRARE(c) == '\n' |
358 | && (BmPREVIOUS(c) == SvCUR(c) - 1) |
359 | && SvTAIL(c) )) |
22e551b9 |
360 | s = screaminstr(sv, prog->check_substr, |
c277df42 |
361 | start_shift + (stringarg - strbeg), |
362 | end_shift, &scream_pos, 0); |
a0d0e21e |
363 | else |
364 | s = Nullch; |
c277df42 |
365 | scream_olds = s; |
0a12ae7d |
366 | } |
a0d0e21e |
367 | else |
c277df42 |
368 | s = fbm_instr((unsigned char*)s + start_shift, |
369 | (unsigned char*)strend - end_shift, |
cf93c79d |
370 | prog->check_substr, PL_multiline ? FBMrf_MULTILINE : 0); |
a0d0e21e |
371 | if (!s) { |
c277df42 |
372 | ++BmUSEFUL(prog->check_substr); /* hooray */ |
a0d0e21e |
373 | goto phooey; /* not present */ |
a0ed51b3 |
374 | } |
375 | else if (s - stringarg > prog->check_offset_max && |
376 | (UTF |
dfe13c55 |
377 | ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) && t >= stringarg) |
a0ed51b3 |
378 | : (t = s - prog->check_offset_max) != 0 |
379 | ) |
380 | ) |
381 | { |
c277df42 |
382 | ++BmUSEFUL(prog->check_substr); /* hooray/2 */ |
a0ed51b3 |
383 | s = t; |
384 | } |
385 | else if (!(prog->reganch & ROPT_NAUGHTY) |
c277df42 |
386 | && --BmUSEFUL(prog->check_substr) < 0 |
387 | && prog->check_substr == prog->float_substr) { /* boo */ |
388 | SvREFCNT_dec(prog->check_substr); |
389 | prog->check_substr = Nullsv; /* disable */ |
390 | prog->float_substr = Nullsv; /* clear */ |
a0d0e21e |
391 | s = startpos; |
a0ed51b3 |
392 | } |
393 | else |
394 | s = startpos; |
a0d0e21e |
395 | } |
a687059c |
396 | |
35ef4773 |
397 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
398 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
8d300b32 |
399 | "%sMatching%s `%s%.60s%s%s' against `%s%.*s%s%s'\n", |
400 | PL_colors[4],PL_colors[5],PL_colors[0], |
401 | prog->precomp, |
402 | PL_colors[1], |
c277df42 |
403 | (strlen(prog->precomp) > 60 ? "..." : ""), |
8d300b32 |
404 | PL_colors[0], |
c277df42 |
405 | (strend - startpos > 60 ? 60 : strend - startpos), |
8d300b32 |
406 | startpos, PL_colors[1], |
c277df42 |
407 | (strend - startpos > 60 ? "..." : "")) |
408 | ); |
409 | |
22e551b9 |
410 | if (prog->reganch & ROPT_GPOS_SEEN) { |
411 | MAGIC *mg; |
22e551b9 |
412 | |
ad94a511 |
413 | if (!(flags & REXEC_IGNOREPOS) && sv && SvTYPE(sv) >= SVt_PVMG |
414 | && SvMAGIC(sv) && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) |
415 | PL_reg_ganch = strbeg + mg->mg_len; |
416 | else |
417 | PL_reg_ganch = startpos; |
22e551b9 |
418 | } |
419 | |
a0d0e21e |
420 | /* Simplest case: anchored match need be tried only once. */ |
774d564b |
421 | /* [unless only anchor is BOL and multiline is set] */ |
22e551b9 |
422 | if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { |
a0d0e21e |
423 | if (regtry(prog, startpos)) |
424 | goto got_it; |
22e551b9 |
425 | else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT) |
426 | || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ |
774d564b |
427 | { |
a0d0e21e |
428 | if (minlen) |
429 | dontbother = minlen - 1; |
dfe13c55 |
430 | strend = HOPc(strend, -dontbother); |
a0d0e21e |
431 | /* for multiline we only have to try after newlines */ |
432 | if (s > startpos) |
433 | s--; |
434 | while (s < strend) { |
6f06b55f |
435 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ |
a0d0e21e |
436 | if (s < strend && regtry(prog, s)) |
437 | goto got_it; |
438 | } |
35c8bce7 |
439 | } |
35c8bce7 |
440 | } |
a0d0e21e |
441 | goto phooey; |
22e551b9 |
442 | } else if (prog->reganch & ROPT_ANCH_GPOS) { |
443 | if (regtry(prog, PL_reg_ganch)) |
444 | goto got_it; |
445 | goto phooey; |
a0d0e21e |
446 | } |
35c8bce7 |
447 | |
a0d0e21e |
448 | /* Messy cases: unanchored match. */ |
c277df42 |
449 | if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { |
450 | /* we have /x+whatever/ */ |
451 | /* it must be a one character string */ |
452 | char ch = SvPVX(prog->anchored_substr)[0]; |
a0ed51b3 |
453 | if (UTF) { |
454 | while (s < strend) { |
455 | if (*s == ch) { |
456 | if (regtry(prog, s)) goto got_it; |
457 | s += UTF8SKIP(s); |
458 | while (s < strend && *s == ch) |
459 | s += UTF8SKIP(s); |
460 | } |
461 | s += UTF8SKIP(s); |
462 | } |
463 | } |
464 | else { |
465 | while (s < strend) { |
466 | if (*s == ch) { |
467 | if (regtry(prog, s)) goto got_it; |
c277df42 |
468 | s++; |
a0ed51b3 |
469 | while (s < strend && *s == ch) |
470 | s++; |
471 | } |
472 | s++; |
a0d0e21e |
473 | } |
a687059c |
474 | } |
c277df42 |
475 | } |
476 | /*SUPPRESS 560*/ |
477 | else if (prog->anchored_substr != Nullsv |
478 | || (prog->float_substr != Nullsv |
479 | && prog->float_max_offset < strend - s)) { |
480 | SV *must = prog->anchored_substr |
481 | ? prog->anchored_substr : prog->float_substr; |
482 | I32 back_max = |
483 | prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; |
484 | I32 back_min = |
485 | prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; |
486 | I32 delta = back_max - back_min; |
cf93c79d |
487 | char *last = HOPc(strend, /* Cannot start after this */ |
73c4f7a1 |
488 | -(I32)(CHR_SVLEN(must) |
489 | - (SvTAIL(must) != 0) + back_min)); |
a0ed51b3 |
490 | char *last1; /* Last position checked before */ |
491 | |
492 | if (s > PL_bostr) |
dfe13c55 |
493 | last1 = HOPc(s, -1); |
a0ed51b3 |
494 | else |
495 | last1 = s - 1; /* bogus */ |
c277df42 |
496 | |
497 | /* XXXX check_substr already used to find `s', can optimize if |
498 | check_substr==must. */ |
499 | scream_pos = -1; |
500 | dontbother = end_shift; |
dfe13c55 |
501 | strend = HOPc(strend, -dontbother); |
c277df42 |
502 | while ( (s <= last) && |
22e551b9 |
503 | ((flags & REXEC_SCREAM) |
504 | ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg, |
c277df42 |
505 | end_shift, &scream_pos, 0)) |
a0ed51b3 |
506 | : (s = fbm_instr((unsigned char*)HOP(s, back_min), |
cf93c79d |
507 | (unsigned char*)strend, must, |
508 | PL_multiline ? FBMrf_MULTILINE : 0))) ) { |
dfe13c55 |
509 | if (HOPc(s, -back_max) > last1) { |
510 | last1 = HOPc(s, -back_min); |
511 | s = HOPc(s, -back_max); |
a0ed51b3 |
512 | } |
513 | else { |
dfe13c55 |
514 | char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
c277df42 |
515 | |
dfe13c55 |
516 | last1 = HOPc(s, -back_min); |
c277df42 |
517 | s = t; |
a0d0e21e |
518 | } |
a0ed51b3 |
519 | if (UTF) { |
520 | while (s <= last1) { |
521 | if (regtry(prog, s)) |
522 | goto got_it; |
523 | s += UTF8SKIP(s); |
524 | } |
525 | } |
526 | else { |
527 | while (s <= last1) { |
528 | if (regtry(prog, s)) |
529 | goto got_it; |
530 | s++; |
531 | } |
a0d0e21e |
532 | } |
533 | } |
534 | goto phooey; |
a0ed51b3 |
535 | } |
536 | else if (c = prog->regstclass) { |
a0d0e21e |
537 | I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
a0ed51b3 |
538 | char *cc; |
a687059c |
539 | |
a0d0e21e |
540 | if (minlen) |
541 | dontbother = minlen - 1; |
dfe13c55 |
542 | strend = HOPc(strend, -dontbother); /* don't bother with what can't match */ |
a0d0e21e |
543 | tmp = 1; |
544 | /* We know what class it must start with. */ |
545 | switch (OP(c)) { |
a0ed51b3 |
546 | case ANYOFUTF8: |
547 | cc = (char *) OPERAND(c); |
548 | while (s < strend) { |
549 | if (REGINCLASSUTF8(c, (U8*)s)) { |
550 | if (tmp && regtry(prog, s)) |
551 | goto got_it; |
552 | else |
553 | tmp = doevery; |
554 | } |
555 | else |
556 | tmp = 1; |
557 | s += UTF8SKIP(s); |
558 | } |
559 | break; |
a0d0e21e |
560 | case ANYOF: |
a0ed51b3 |
561 | cc = (char *) OPERAND(c); |
a0d0e21e |
562 | while (s < strend) { |
a0ed51b3 |
563 | if (REGINCLASS(cc, *s)) { |
a0d0e21e |
564 | if (tmp && regtry(prog, s)) |
565 | goto got_it; |
566 | else |
567 | tmp = doevery; |
a687059c |
568 | } |
a0d0e21e |
569 | else |
570 | tmp = 1; |
571 | s++; |
572 | } |
573 | break; |
bbce6d69 |
574 | case BOUNDL: |
3280af22 |
575 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
576 | /* FALL THROUGH */ |
a0d0e21e |
577 | case BOUND: |
a0ed51b3 |
578 | if (minlen) { |
579 | dontbother++; |
580 | strend -= 1; |
581 | } |
3280af22 |
582 | tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; |
95bac841 |
583 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
584 | while (s < strend) { |
95bac841 |
585 | if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { |
a0d0e21e |
586 | tmp = !tmp; |
587 | if (regtry(prog, s)) |
588 | goto got_it; |
a687059c |
589 | } |
a0d0e21e |
590 | s++; |
591 | } |
592 | if ((minlen || tmp) && regtry(prog,s)) |
593 | goto got_it; |
594 | break; |
a0ed51b3 |
595 | case BOUNDLUTF8: |
596 | PL_reg_flags |= RF_tainted; |
597 | /* FALL THROUGH */ |
598 | case BOUNDUTF8: |
599 | if (minlen) { |
600 | dontbother++; |
dfe13c55 |
601 | strend = reghop_c(strend, -1); |
a0ed51b3 |
602 | } |
dfe13c55 |
603 | tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; |
a0ed51b3 |
604 | tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); |
605 | while (s < strend) { |
dfe13c55 |
606 | if (tmp == !(OP(c) == BOUND ? |
607 | swash_fetch(PL_utf8_alnum, (U8*)s) : |
608 | isALNUM_LC_utf8((U8*)s))) |
609 | { |
a0ed51b3 |
610 | tmp = !tmp; |
611 | if (regtry(prog, s)) |
612 | goto got_it; |
613 | } |
614 | s += UTF8SKIP(s); |
615 | } |
616 | if ((minlen || tmp) && regtry(prog,s)) |
617 | goto got_it; |
618 | break; |
bbce6d69 |
619 | case NBOUNDL: |
3280af22 |
620 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
621 | /* FALL THROUGH */ |
a0d0e21e |
622 | case NBOUND: |
a0ed51b3 |
623 | if (minlen) { |
624 | dontbother++; |
625 | strend -= 1; |
626 | } |
3280af22 |
627 | tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; |
95bac841 |
628 | tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
629 | while (s < strend) { |
95bac841 |
630 | if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) |
a0d0e21e |
631 | tmp = !tmp; |
632 | else if (regtry(prog, s)) |
633 | goto got_it; |
634 | s++; |
635 | } |
636 | if ((minlen || !tmp) && regtry(prog,s)) |
637 | goto got_it; |
638 | break; |
a0ed51b3 |
639 | case NBOUNDLUTF8: |
640 | PL_reg_flags |= RF_tainted; |
641 | /* FALL THROUGH */ |
642 | case NBOUNDUTF8: |
643 | if (minlen) { |
644 | dontbother++; |
dfe13c55 |
645 | strend = reghop_c(strend, -1); |
a0ed51b3 |
646 | } |
dfe13c55 |
647 | tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; |
a0ed51b3 |
648 | tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); |
649 | while (s < strend) { |
dfe13c55 |
650 | if (tmp == !(OP(c) == NBOUND ? |
651 | swash_fetch(PL_utf8_alnum, (U8*)s) : |
652 | isALNUM_LC_utf8((U8*)s))) |
a0ed51b3 |
653 | tmp = !tmp; |
654 | else if (regtry(prog, s)) |
655 | goto got_it; |
656 | s += UTF8SKIP(s); |
657 | } |
658 | if ((minlen || !tmp) && regtry(prog,s)) |
659 | goto got_it; |
660 | break; |
a0d0e21e |
661 | case ALNUM: |
662 | while (s < strend) { |
bbce6d69 |
663 | if (isALNUM(*s)) { |
664 | if (tmp && regtry(prog, s)) |
665 | goto got_it; |
666 | else |
667 | tmp = doevery; |
668 | } |
669 | else |
670 | tmp = 1; |
671 | s++; |
672 | } |
673 | break; |
a0ed51b3 |
674 | case ALNUMUTF8: |
675 | while (s < strend) { |
dfe13c55 |
676 | if (swash_fetch(PL_utf8_alnum, (U8*)s)) { |
a0ed51b3 |
677 | if (tmp && regtry(prog, s)) |
678 | goto got_it; |
679 | else |
680 | tmp = doevery; |
681 | } |
682 | else |
683 | tmp = 1; |
684 | s += UTF8SKIP(s); |
685 | } |
686 | break; |
bbce6d69 |
687 | case ALNUML: |
3280af22 |
688 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
689 | while (s < strend) { |
690 | if (isALNUM_LC(*s)) { |
a0d0e21e |
691 | if (tmp && regtry(prog, s)) |
692 | goto got_it; |
a687059c |
693 | else |
a0d0e21e |
694 | tmp = doevery; |
695 | } |
696 | else |
697 | tmp = 1; |
698 | s++; |
699 | } |
700 | break; |
a0ed51b3 |
701 | case ALNUMLUTF8: |
702 | PL_reg_flags |= RF_tainted; |
703 | while (s < strend) { |
dfe13c55 |
704 | if (isALNUM_LC_utf8((U8*)s)) { |
a0ed51b3 |
705 | if (tmp && regtry(prog, s)) |
706 | goto got_it; |
707 | else |
708 | tmp = doevery; |
709 | } |
710 | else |
711 | tmp = 1; |
712 | s += UTF8SKIP(s); |
713 | } |
714 | break; |
a0d0e21e |
715 | case NALNUM: |
716 | while (s < strend) { |
bbce6d69 |
717 | if (!isALNUM(*s)) { |
718 | if (tmp && regtry(prog, s)) |
719 | goto got_it; |
720 | else |
721 | tmp = doevery; |
722 | } |
723 | else |
724 | tmp = 1; |
725 | s++; |
726 | } |
727 | break; |
a0ed51b3 |
728 | case NALNUMUTF8: |
729 | while (s < strend) { |
dfe13c55 |
730 | if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { |
a0ed51b3 |
731 | if (tmp && regtry(prog, s)) |
732 | goto got_it; |
733 | else |
734 | tmp = doevery; |
735 | } |
736 | else |
737 | tmp = 1; |
738 | s += UTF8SKIP(s); |
739 | } |
740 | break; |
bbce6d69 |
741 | case NALNUML: |
3280af22 |
742 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
743 | while (s < strend) { |
744 | if (!isALNUM_LC(*s)) { |
a0d0e21e |
745 | if (tmp && regtry(prog, s)) |
746 | goto got_it; |
a687059c |
747 | else |
a0d0e21e |
748 | tmp = doevery; |
a687059c |
749 | } |
a0d0e21e |
750 | else |
751 | tmp = 1; |
752 | s++; |
753 | } |
754 | break; |
a0ed51b3 |
755 | case NALNUMLUTF8: |
756 | PL_reg_flags |= RF_tainted; |
757 | while (s < strend) { |
dfe13c55 |
758 | if (!isALNUM_LC_utf8((U8*)s)) { |
a0ed51b3 |
759 | if (tmp && regtry(prog, s)) |
760 | goto got_it; |
761 | else |
762 | tmp = doevery; |
763 | } |
764 | else |
765 | tmp = 1; |
766 | s += UTF8SKIP(s); |
767 | } |
768 | break; |
a0d0e21e |
769 | case SPACE: |
770 | while (s < strend) { |
771 | if (isSPACE(*s)) { |
772 | if (tmp && regtry(prog, s)) |
773 | goto got_it; |
774 | else |
775 | tmp = doevery; |
2304df62 |
776 | } |
a0d0e21e |
777 | else |
778 | tmp = 1; |
779 | s++; |
780 | } |
781 | break; |
a0ed51b3 |
782 | case SPACEUTF8: |
783 | while (s < strend) { |
dfe13c55 |
784 | if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { |
a0ed51b3 |
785 | if (tmp && regtry(prog, s)) |
786 | goto got_it; |
787 | else |
788 | tmp = doevery; |
789 | } |
790 | else |
791 | tmp = 1; |
792 | s += UTF8SKIP(s); |
793 | } |
794 | break; |
bbce6d69 |
795 | case SPACEL: |
3280af22 |
796 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
797 | while (s < strend) { |
798 | if (isSPACE_LC(*s)) { |
799 | if (tmp && regtry(prog, s)) |
800 | goto got_it; |
801 | else |
802 | tmp = doevery; |
803 | } |
804 | else |
805 | tmp = 1; |
806 | s++; |
807 | } |
808 | break; |
a0ed51b3 |
809 | case SPACELUTF8: |
810 | PL_reg_flags |= RF_tainted; |
811 | while (s < strend) { |
dfe13c55 |
812 | if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) { |
a0ed51b3 |
813 | if (tmp && regtry(prog, s)) |
814 | goto got_it; |
815 | else |
816 | tmp = doevery; |
817 | } |
818 | else |
819 | tmp = 1; |
820 | s += UTF8SKIP(s); |
821 | } |
822 | break; |
a0d0e21e |
823 | case NSPACE: |
824 | while (s < strend) { |
825 | if (!isSPACE(*s)) { |
826 | if (tmp && regtry(prog, s)) |
827 | goto got_it; |
828 | else |
829 | tmp = doevery; |
a687059c |
830 | } |
a0d0e21e |
831 | else |
832 | tmp = 1; |
833 | s++; |
834 | } |
835 | break; |
a0ed51b3 |
836 | case NSPACEUTF8: |
837 | while (s < strend) { |
dfe13c55 |
838 | if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { |
a0ed51b3 |
839 | if (tmp && regtry(prog, s)) |
840 | goto got_it; |
841 | else |
842 | tmp = doevery; |
843 | } |
844 | else |
845 | tmp = 1; |
846 | s += UTF8SKIP(s); |
847 | } |
848 | break; |
bbce6d69 |
849 | case NSPACEL: |
3280af22 |
850 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
851 | while (s < strend) { |
852 | if (!isSPACE_LC(*s)) { |
853 | if (tmp && regtry(prog, s)) |
854 | goto got_it; |
855 | else |
856 | tmp = doevery; |
857 | } |
858 | else |
859 | tmp = 1; |
860 | s++; |
861 | } |
862 | break; |
a0ed51b3 |
863 | case NSPACELUTF8: |
864 | PL_reg_flags |= RF_tainted; |
865 | while (s < strend) { |
dfe13c55 |
866 | if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) { |
a0ed51b3 |
867 | if (tmp && regtry(prog, s)) |
868 | goto got_it; |
869 | else |
870 | tmp = doevery; |
871 | } |
872 | else |
873 | tmp = 1; |
874 | s += UTF8SKIP(s); |
875 | } |
876 | break; |
a0d0e21e |
877 | case DIGIT: |
878 | while (s < strend) { |
879 | if (isDIGIT(*s)) { |
880 | if (tmp && regtry(prog, s)) |
881 | goto got_it; |
882 | else |
883 | tmp = doevery; |
2b69d0c2 |
884 | } |
a0d0e21e |
885 | else |
886 | tmp = 1; |
887 | s++; |
888 | } |
889 | break; |
a0ed51b3 |
890 | case DIGITUTF8: |
891 | while (s < strend) { |
dfe13c55 |
892 | if (swash_fetch(PL_utf8_digit,(U8*)s)) { |
a0ed51b3 |
893 | if (tmp && regtry(prog, s)) |
894 | goto got_it; |
895 | else |
896 | tmp = doevery; |
897 | } |
898 | else |
899 | tmp = 1; |
900 | s += UTF8SKIP(s); |
901 | } |
902 | break; |
a0d0e21e |
903 | case NDIGIT: |
904 | while (s < strend) { |
905 | if (!isDIGIT(*s)) { |
906 | if (tmp && regtry(prog, s)) |
907 | goto got_it; |
908 | else |
909 | tmp = doevery; |
a687059c |
910 | } |
a0d0e21e |
911 | else |
912 | tmp = 1; |
913 | s++; |
914 | } |
915 | break; |
a0ed51b3 |
916 | case NDIGITUTF8: |
917 | while (s < strend) { |
dfe13c55 |
918 | if (!swash_fetch(PL_utf8_digit,(U8*)s)) { |
a0ed51b3 |
919 | if (tmp && regtry(prog, s)) |
920 | goto got_it; |
921 | else |
922 | tmp = doevery; |
923 | } |
924 | else |
925 | tmp = 1; |
926 | s += UTF8SKIP(s); |
927 | } |
928 | break; |
a687059c |
929 | } |
a0d0e21e |
930 | } |
931 | else { |
c277df42 |
932 | dontbother = 0; |
933 | if (prog->float_substr != Nullsv) { /* Trim the end. */ |
934 | char *last; |
935 | I32 oldpos = scream_pos; |
936 | |
22e551b9 |
937 | if (flags & REXEC_SCREAM) { |
938 | last = screaminstr(sv, prog->float_substr, s - strbeg, |
c277df42 |
939 | end_shift, &scream_pos, 1); /* last one */ |
cf93c79d |
940 | if (!last) |
c277df42 |
941 | last = scream_olds; /* Only one occurence. */ |
a0ed51b3 |
942 | } |
943 | else { |
c277df42 |
944 | STRLEN len; |
945 | char *little = SvPV(prog->float_substr, len); |
cf93c79d |
946 | |
947 | if (SvTAIL(prog->float_substr)) { |
948 | if (memEQ(strend - len + 1, little, len - 1)) |
949 | last = strend - len + 1; |
950 | else if (!PL_multiline) |
951 | last = memEQ(strend - len, little, len) |
952 | ? strend - len : Nullch; |
953 | else |
954 | goto find_last; |
955 | } else { |
956 | find_last: |
957 | if (len) |
958 | last = rninstr(s, strend, little, little + len); |
959 | else |
960 | last = strend; /* matching `$' */ |
961 | } |
c277df42 |
962 | } |
963 | if (last == NULL) goto phooey; /* Should not happen! */ |
19b4f81a |
964 | dontbother = strend - last + prog->float_min_offset; |
c277df42 |
965 | } |
966 | if (minlen && (dontbother < minlen)) |
a0d0e21e |
967 | dontbother = minlen - 1; |
a0ed51b3 |
968 | strend -= dontbother; /* this one's always in bytes! */ |
a0d0e21e |
969 | /* We don't know much -- general case. */ |
a0ed51b3 |
970 | if (UTF) { |
971 | for (;;) { |
84df6dba |
972 | if (regtry(prog, s)) |
a0ed51b3 |
973 | goto got_it; |
a0ed51b3 |
974 | if (s >= strend) |
975 | break; |
976 | s += UTF8SKIP(s); |
977 | }; |
978 | } |
979 | else { |
980 | do { |
981 | if (regtry(prog, s)) |
982 | goto got_it; |
983 | } while (s++ < strend); |
984 | } |
a0d0e21e |
985 | } |
986 | |
987 | /* Failure. */ |
988 | goto phooey; |
a687059c |
989 | |
a0d0e21e |
990 | got_it: |
3280af22 |
991 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); |
5f05dabc |
992 | |
5c5e4c24 |
993 | if (PL_reg_eval_set) { |
994 | /* Preserve the current value of $^R */ |
995 | if (oreplsv != GvSV(PL_replgv)) |
996 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is |
997 | restored, the value remains |
998 | the same. */ |
9661b544 |
999 | restore_pos(0); |
5c5e4c24 |
1000 | } |
cf93c79d |
1001 | |
1002 | /* make sure $`, $&, $', and $digit will work later */ |
1003 | if ( !(flags & REXEC_NOT_FIRST) ) { |
1004 | if (RX_MATCH_COPIED(prog)) { |
1005 | Safefree(prog->subbeg); |
1006 | RX_MATCH_COPIED_off(prog); |
1007 | } |
1008 | if (flags & REXEC_COPY_STR) { |
1009 | I32 i = PL_regeol - startpos + (stringarg - strbeg); |
1010 | |
1011 | s = savepvn(strbeg, i); |
1012 | prog->subbeg = s; |
1013 | prog->sublen = i; |
1014 | RX_MATCH_COPIED_on(prog); |
1015 | } |
1016 | else { |
1017 | prog->subbeg = strbeg; |
1018 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ |
1019 | } |
1020 | } |
5c5e4c24 |
1021 | |
a0d0e21e |
1022 | return 1; |
1023 | |
1024 | phooey: |
9661b544 |
1025 | if (PL_reg_eval_set) |
1026 | restore_pos(0); |
a0d0e21e |
1027 | return 0; |
a687059c |
1028 | } |
1029 | |
1030 | /* |
1031 | - regtry - try match at specific point |
1032 | */ |
76e3520e |
1033 | STATIC I32 /* 0 failure, 1 success */ |
cea2e8a9 |
1034 | S_regtry(pTHX_ regexp *prog, char *startpos) |
a687059c |
1035 | { |
c277df42 |
1036 | dTHR; |
a0d0e21e |
1037 | register I32 i; |
cf93c79d |
1038 | register I32 *sp; |
1039 | register I32 *ep; |
c277df42 |
1040 | CHECKPOINT lastcp; |
a0d0e21e |
1041 | |
3280af22 |
1042 | if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { |
9661b544 |
1043 | MAGIC *mg; |
1044 | |
3280af22 |
1045 | PL_reg_eval_set = RS_init; |
ce862d02 |
1046 | DEBUG_r(DEBUG_s( |
c3464db5 |
1047 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n", |
3280af22 |
1048 | PL_stack_sp - PL_stack_base); |
ce862d02 |
1049 | )); |
1050 | SAVEINT(cxstack[cxstack_ix].blk_oldsp); |
3280af22 |
1051 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
ce862d02 |
1052 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ |
1053 | SAVETMPS; |
1054 | /* Apparently this is not needed, judging by wantarray. */ |
1055 | /* SAVEINT(cxstack[cxstack_ix].blk_gimme); |
1056 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
9661b544 |
1057 | |
1058 | if (PL_reg_sv) { |
1059 | /* Make $_ available to executed code. */ |
127ad2b7 |
1060 | if (PL_reg_sv != DEFSV) { |
1061 | /* SAVE_DEFSV does *not* suffice here for USE_THREADS */ |
1062 | SAVESPTR(DEFSV); |
1063 | DEFSV = PL_reg_sv; |
9661b544 |
1064 | } |
1065 | |
1066 | if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) |
1067 | && (mg = mg_find(PL_reg_sv, 'g')))) { |
1068 | /* prepare for quick setting of pos */ |
1069 | sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0); |
1070 | mg = mg_find(PL_reg_sv, 'g'); |
1071 | mg->mg_len = -1; |
1072 | } |
1073 | PL_reg_magic = mg; |
1074 | PL_reg_oldpos = mg->mg_len; |
cea2e8a9 |
1075 | SAVEDESTRUCTOR(S_restore_pos, 0); |
9661b544 |
1076 | } |
5c5e4c24 |
1077 | if (!PL_reg_curpm) |
1078 | New(22,PL_reg_curpm, 1, PMOP); |
1079 | PL_reg_curpm->op_pmregexp = prog; |
1080 | PL_reg_oldcurpm = PL_curpm; |
1081 | PL_curpm = PL_reg_curpm; |
cf93c79d |
1082 | if (RX_MATCH_COPIED(prog)) { |
1083 | /* Here is a serious problem: we cannot rewrite subbeg, |
1084 | since it may be needed if this match fails. Thus |
1085 | $` inside (?{}) could fail... */ |
1086 | PL_reg_oldsaved = prog->subbeg; |
1087 | PL_reg_oldsavedlen = prog->sublen; |
1088 | RX_MATCH_COPIED_off(prog); |
1089 | } |
1090 | else |
1091 | PL_reg_oldsaved = Nullch; |
5c5e4c24 |
1092 | prog->subbeg = PL_bostr; |
cf93c79d |
1093 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ |
ce862d02 |
1094 | } |
cf93c79d |
1095 | prog->startp[0] = startpos - PL_bostr; |
3280af22 |
1096 | PL_reginput = startpos; |
1097 | PL_regstartp = prog->startp; |
1098 | PL_regendp = prog->endp; |
1099 | PL_reglastparen = &prog->lastparen; |
a0d0e21e |
1100 | prog->lastparen = 0; |
3280af22 |
1101 | PL_regsize = 0; |
364723c2 |
1102 | DEBUG_r(PL_reg_starttry = startpos); |
3280af22 |
1103 | if (PL_reg_start_tmpl <= prog->nparens) { |
1104 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; |
1105 | if(PL_reg_start_tmp) |
1106 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
c277df42 |
1107 | else |
3280af22 |
1108 | New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
c277df42 |
1109 | } |
a0d0e21e |
1110 | |
5c5e4c24 |
1111 | /* XXXX What this code is doing here?!!! There should be no need |
1112 | to do this again and again, PL_reglastparen should take care of |
1113 | this! */ |
a0d0e21e |
1114 | sp = prog->startp; |
1115 | ep = prog->endp; |
1116 | if (prog->nparens) { |
5c5e4c24 |
1117 | for (i = prog->nparens; i >= 1; i--) { |
cf93c79d |
1118 | *++sp = -1; |
1119 | *++ep = -1; |
a687059c |
1120 | } |
a0d0e21e |
1121 | } |
c277df42 |
1122 | REGCP_SET; |
7e5428c5 |
1123 | if (regmatch(prog->program + 1)) { |
cf93c79d |
1124 | prog->endp[0] = PL_reginput - PL_bostr; |
a0d0e21e |
1125 | return 1; |
1126 | } |
c277df42 |
1127 | REGCP_UNWIND; |
1128 | return 0; |
a687059c |
1129 | } |
1130 | |
1131 | /* |
1132 | - regmatch - main matching routine |
1133 | * |
1134 | * Conceptually the strategy is simple: check to see whether the current |
1135 | * node matches, call self recursively to see whether the rest matches, |
1136 | * and then act accordingly. In practice we make some effort to avoid |
1137 | * recursion, in particular by going through "ordinary" nodes (that don't |
1138 | * need to know whether the rest of the match failed) by a loop instead of |
1139 | * by recursion. |
1140 | */ |
1141 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
1142 | * maybe save a little bit of pushing and popping on the stack. It also takes |
1143 | * advantage of machines that use a register save mask on subroutine entry. |
1144 | */ |
76e3520e |
1145 | STATIC I32 /* 0 failure, 1 success */ |
cea2e8a9 |
1146 | S_regmatch(pTHX_ regnode *prog) |
a687059c |
1147 | { |
c277df42 |
1148 | dTHR; |
1149 | register regnode *scan; /* Current node. */ |
1150 | regnode *next; /* Next node. */ |
1151 | regnode *inner; /* Next node in internal branch. */ |
c3464db5 |
1152 | register I32 nextchr; /* renamed nextchr - nextchar colides with |
1153 | function of same name */ |
a0d0e21e |
1154 | register I32 n; /* no or next */ |
1155 | register I32 ln; /* len or last */ |
1156 | register char *s; /* operand or save */ |
3280af22 |
1157 | register char *locinput = PL_reginput; |
c277df42 |
1158 | register I32 c1, c2, paren; /* case fold search, parenth */ |
1159 | int minmod = 0, sw = 0, logical = 0; |
4633a7c4 |
1160 | #ifdef DEBUGGING |
3280af22 |
1161 | PL_regindent++; |
4633a7c4 |
1162 | #endif |
a0d0e21e |
1163 | |
a0ed51b3 |
1164 | /* Note that nextchr is a byte even in UTF */ |
76e3520e |
1165 | nextchr = UCHARAT(locinput); |
a0d0e21e |
1166 | scan = prog; |
1167 | while (scan != NULL) { |
c277df42 |
1168 | #define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO) |
a687059c |
1169 | #ifdef DEBUGGING |
c277df42 |
1170 | # define sayYES goto yes |
1171 | # define sayNO goto no |
1172 | # define saySAME(x) if (x) goto yes; else goto no |
1173 | # define REPORT_CODE_OFF 24 |
4633a7c4 |
1174 | #else |
c277df42 |
1175 | # define sayYES return 1 |
1176 | # define sayNO return 0 |
1177 | # define saySAME(x) return x |
a687059c |
1178 | #endif |
c277df42 |
1179 | DEBUG_r( { |
1180 | SV *prop = sv_newmortal(); |
3280af22 |
1181 | int docolor = *PL_colors[0]; |
c277df42 |
1182 | int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
3280af22 |
1183 | int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput); |
beed8111 |
1184 | /* The part of the string before starttry has one color |
1185 | (pref0_len chars), between starttry and current |
1186 | position another one (pref_len - pref0_len chars), |
1187 | after the current position the third one. |
1188 | We assume that pref0_len <= pref_len, otherwise we |
1189 | decrease pref0_len. */ |
3280af22 |
1190 | int pref_len = (locinput - PL_bostr > (5 + taill) - l |
1191 | ? (5 + taill) - l : locinput - PL_bostr); |
364723c2 |
1192 | int pref0_len = pref_len - (locinput - PL_reg_starttry); |
c277df42 |
1193 | |
3280af22 |
1194 | if (l + pref_len < (5 + taill) && l < PL_regeol - locinput) |
1195 | l = ( PL_regeol - locinput > (5 + taill) - pref_len |
1196 | ? (5 + taill) - pref_len : PL_regeol - locinput); |
8d300b32 |
1197 | if (pref0_len < 0) |
1198 | pref0_len = 0; |
beed8111 |
1199 | if (pref0_len > pref_len) |
1200 | pref0_len = pref_len; |
c277df42 |
1201 | regprop(prop, scan); |
1202 | PerlIO_printf(Perl_debug_log, |
8d300b32 |
1203 | "%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n", |
3280af22 |
1204 | locinput - PL_bostr, |
8d300b32 |
1205 | PL_colors[4], pref0_len, |
1206 | locinput - pref_len, PL_colors[5], |
1207 | PL_colors[2], pref_len - pref0_len, |
1208 | locinput - pref_len + pref0_len, PL_colors[3], |
c277df42 |
1209 | (docolor ? "" : "> <"), |
3280af22 |
1210 | PL_colors[0], l, locinput, PL_colors[1], |
c277df42 |
1211 | 15 - l - pref_len + 1, |
1212 | "", |
3280af22 |
1213 | scan - PL_regprogram, PL_regindent*2, "", |
c277df42 |
1214 | SvPVX(prop)); |
1215 | } ); |
a687059c |
1216 | |
c277df42 |
1217 | next = scan + NEXT_OFF(scan); |
a0d0e21e |
1218 | if (next == scan) |
1219 | next = NULL; |
a687059c |
1220 | |
a0d0e21e |
1221 | switch (OP(scan)) { |
1222 | case BOL: |
3280af22 |
1223 | if (locinput == PL_bostr |
1224 | ? PL_regprev == '\n' |
1225 | : (PL_multiline && |
1226 | (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) |
a0d0e21e |
1227 | { |
a0ed51b3 |
1228 | /* regtill = regbol; */ |
a0d0e21e |
1229 | break; |
1230 | } |
4633a7c4 |
1231 | sayNO; |
a0d0e21e |
1232 | case MBOL: |
3280af22 |
1233 | if (locinput == PL_bostr |
1234 | ? PL_regprev == '\n' |
1235 | : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) |
a0d0e21e |
1236 | { |
1237 | break; |
1238 | } |
4633a7c4 |
1239 | sayNO; |
a0d0e21e |
1240 | case SBOL: |
3280af22 |
1241 | if (locinput == PL_regbol && PL_regprev == '\n') |
a0d0e21e |
1242 | break; |
4633a7c4 |
1243 | sayNO; |
774d564b |
1244 | case GPOS: |
22e551b9 |
1245 | if (locinput == PL_reg_ganch) |
a0d0e21e |
1246 | break; |
4633a7c4 |
1247 | sayNO; |
a0d0e21e |
1248 | case EOL: |
3280af22 |
1249 | if (PL_multiline) |
a0d0e21e |
1250 | goto meol; |
1251 | else |
1252 | goto seol; |
1253 | case MEOL: |
1254 | meol: |
3280af22 |
1255 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
4633a7c4 |
1256 | sayNO; |
a0d0e21e |
1257 | break; |
1258 | case SEOL: |
1259 | seol: |
3280af22 |
1260 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
4633a7c4 |
1261 | sayNO; |
3280af22 |
1262 | if (PL_regeol - locinput > 1) |
4633a7c4 |
1263 | sayNO; |
a0d0e21e |
1264 | break; |
b85d18e9 |
1265 | case EOS: |
3280af22 |
1266 | if (PL_regeol != locinput) |
b85d18e9 |
1267 | sayNO; |
1268 | break; |
a0ed51b3 |
1269 | case SANYUTF8: |
1270 | if (nextchr & 0x80) { |
6f06b55f |
1271 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1272 | if (locinput > PL_regeol) |
1273 | sayNO; |
1274 | nextchr = UCHARAT(locinput); |
1275 | break; |
1276 | } |
1277 | if (!nextchr && locinput >= PL_regeol) |
1278 | sayNO; |
1279 | nextchr = UCHARAT(++locinput); |
1280 | break; |
a0d0e21e |
1281 | case SANY: |
3280af22 |
1282 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1283 | sayNO; |
76e3520e |
1284 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1285 | break; |
a0ed51b3 |
1286 | case ANYUTF8: |
1287 | if (nextchr & 0x80) { |
6f06b55f |
1288 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1289 | if (locinput > PL_regeol) |
1290 | sayNO; |
1291 | nextchr = UCHARAT(locinput); |
1292 | break; |
1293 | } |
1294 | if (!nextchr && locinput >= PL_regeol || nextchr == '\n') |
1295 | sayNO; |
1296 | nextchr = UCHARAT(++locinput); |
1297 | break; |
22c35a8c |
1298 | case REG_ANY: |
3280af22 |
1299 | if (!nextchr && locinput >= PL_regeol || nextchr == '\n') |
4633a7c4 |
1300 | sayNO; |
76e3520e |
1301 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1302 | break; |
bbce6d69 |
1303 | case EXACT: |
161b471a |
1304 | s = (char *) OPERAND(scan); |
c277df42 |
1305 | ln = UCHARAT(s++); |
a0d0e21e |
1306 | /* Inline the first character, for speed. */ |
76e3520e |
1307 | if (UCHARAT(s) != nextchr) |
4633a7c4 |
1308 | sayNO; |
3280af22 |
1309 | if (PL_regeol - locinput < ln) |
4633a7c4 |
1310 | sayNO; |
36477c24 |
1311 | if (ln > 1 && memNE(s, locinput, ln)) |
4633a7c4 |
1312 | sayNO; |
a0d0e21e |
1313 | locinput += ln; |
76e3520e |
1314 | nextchr = UCHARAT(locinput); |
bbce6d69 |
1315 | break; |
1316 | case EXACTFL: |
3280af22 |
1317 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1318 | /* FALL THROUGH */ |
1319 | case EXACTF: |
161b471a |
1320 | s = (char *) OPERAND(scan); |
c277df42 |
1321 | ln = UCHARAT(s++); |
a0ed51b3 |
1322 | |
1323 | if (UTF) { |
1324 | char *l = locinput; |
1325 | char *e = s + ln; |
1326 | c1 = OP(scan) == EXACTF; |
1327 | while (s < e) { |
1328 | if (l >= PL_regeol) |
1329 | sayNO; |
dfe13c55 |
1330 | if (utf8_to_uv((U8*)s, 0) != (c1 ? |
1331 | toLOWER_utf8((U8*)l) : |
1332 | toLOWER_LC_utf8((U8*)l))) |
1333 | { |
a0ed51b3 |
1334 | sayNO; |
dfe13c55 |
1335 | } |
a0ed51b3 |
1336 | s += UTF8SKIP(s); |
1337 | l += UTF8SKIP(l); |
1338 | } |
1339 | locinput = l; |
1340 | nextchr = UCHARAT(locinput); |
1341 | break; |
1342 | } |
1343 | |
bbce6d69 |
1344 | /* Inline the first character, for speed. */ |
76e3520e |
1345 | if (UCHARAT(s) != nextchr && |
bbce6d69 |
1346 | UCHARAT(s) != ((OP(scan) == EXACTF) |
22c35a8c |
1347 | ? PL_fold : PL_fold_locale)[nextchr]) |
bbce6d69 |
1348 | sayNO; |
3280af22 |
1349 | if (PL_regeol - locinput < ln) |
bbce6d69 |
1350 | sayNO; |
5f05dabc |
1351 | if (ln > 1 && (OP(scan) == EXACTF |
1352 | ? ibcmp(s, locinput, ln) |
1353 | : ibcmp_locale(s, locinput, ln))) |
bbce6d69 |
1354 | sayNO; |
1355 | locinput += ln; |
76e3520e |
1356 | nextchr = UCHARAT(locinput); |
a0d0e21e |
1357 | break; |
a0ed51b3 |
1358 | case ANYOFUTF8: |
1359 | s = (char *) OPERAND(scan); |
1360 | if (!REGINCLASSUTF8(scan, (U8*)locinput)) |
1361 | sayNO; |
1362 | if (locinput >= PL_regeol) |
1363 | sayNO; |
6f06b55f |
1364 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1365 | nextchr = UCHARAT(locinput); |
1366 | break; |
a0d0e21e |
1367 | case ANYOF: |
161b471a |
1368 | s = (char *) OPERAND(scan); |
76e3520e |
1369 | if (nextchr < 0) |
1370 | nextchr = UCHARAT(locinput); |
873ef191 |
1371 | if (!REGINCLASS(s, nextchr)) |
4633a7c4 |
1372 | sayNO; |
3280af22 |
1373 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1374 | sayNO; |
76e3520e |
1375 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1376 | break; |
bbce6d69 |
1377 | case ALNUML: |
3280af22 |
1378 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1379 | /* FALL THROUGH */ |
a0d0e21e |
1380 | case ALNUM: |
76e3520e |
1381 | if (!nextchr) |
4633a7c4 |
1382 | sayNO; |
bbce6d69 |
1383 | if (!(OP(scan) == ALNUM |
76e3520e |
1384 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
4633a7c4 |
1385 | sayNO; |
76e3520e |
1386 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1387 | break; |
a0ed51b3 |
1388 | case ALNUMLUTF8: |
1389 | PL_reg_flags |= RF_tainted; |
1390 | /* FALL THROUGH */ |
1391 | case ALNUMUTF8: |
1392 | if (!nextchr) |
1393 | sayNO; |
1394 | if (nextchr & 0x80) { |
1395 | if (!(OP(scan) == ALNUMUTF8 |
dfe13c55 |
1396 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput) |
1397 | : isALNUM_LC_utf8((U8*)locinput))) |
1398 | { |
a0ed51b3 |
1399 | sayNO; |
dfe13c55 |
1400 | } |
6f06b55f |
1401 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1402 | nextchr = UCHARAT(locinput); |
1403 | break; |
1404 | } |
1405 | if (!(OP(scan) == ALNUMUTF8 |
1406 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
1407 | sayNO; |
1408 | nextchr = UCHARAT(++locinput); |
1409 | break; |
bbce6d69 |
1410 | case NALNUML: |
3280af22 |
1411 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1412 | /* FALL THROUGH */ |
a0d0e21e |
1413 | case NALNUM: |
3280af22 |
1414 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1415 | sayNO; |
bbce6d69 |
1416 | if (OP(scan) == NALNUM |
76e3520e |
1417 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
4633a7c4 |
1418 | sayNO; |
76e3520e |
1419 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1420 | break; |
a0ed51b3 |
1421 | case NALNUMLUTF8: |
1422 | PL_reg_flags |= RF_tainted; |
1423 | /* FALL THROUGH */ |
1424 | case NALNUMUTF8: |
1425 | if (!nextchr && locinput >= PL_regeol) |
1426 | sayNO; |
1427 | if (nextchr & 0x80) { |
1428 | if (OP(scan) == NALNUMUTF8 |
dfe13c55 |
1429 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput) |
1430 | : isALNUM_LC_utf8((U8*)locinput)) |
1431 | { |
a0ed51b3 |
1432 | sayNO; |
dfe13c55 |
1433 | } |
6f06b55f |
1434 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1435 | nextchr = UCHARAT(locinput); |
1436 | break; |
1437 | } |
1438 | if (OP(scan) == NALNUMUTF8 |
1439 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
1440 | sayNO; |
1441 | nextchr = UCHARAT(++locinput); |
1442 | break; |
bbce6d69 |
1443 | case BOUNDL: |
1444 | case NBOUNDL: |
3280af22 |
1445 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1446 | /* FALL THROUGH */ |
a0d0e21e |
1447 | case BOUND: |
bbce6d69 |
1448 | case NBOUND: |
1449 | /* was last char in word? */ |
3280af22 |
1450 | ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev; |
bbce6d69 |
1451 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
1452 | ln = isALNUM(ln); |
76e3520e |
1453 | n = isALNUM(nextchr); |
bbce6d69 |
1454 | } |
1455 | else { |
1456 | ln = isALNUM_LC(ln); |
76e3520e |
1457 | n = isALNUM_LC(nextchr); |
bbce6d69 |
1458 | } |
95bac841 |
1459 | if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) |
4633a7c4 |
1460 | sayNO; |
a0d0e21e |
1461 | break; |
a0ed51b3 |
1462 | case BOUNDLUTF8: |
1463 | case NBOUNDLUTF8: |
1464 | PL_reg_flags |= RF_tainted; |
1465 | /* FALL THROUGH */ |
1466 | case BOUNDUTF8: |
1467 | case NBOUNDUTF8: |
1468 | /* was last char in word? */ |
dfe13c55 |
1469 | ln = (locinput != PL_regbol) |
1470 | ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev; |
a0ed51b3 |
1471 | if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) { |
1472 | ln = isALNUM_uni(ln); |
dfe13c55 |
1473 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput); |
a0ed51b3 |
1474 | } |
1475 | else { |
1476 | ln = isALNUM_LC_uni(ln); |
dfe13c55 |
1477 | n = isALNUM_LC_utf8((U8*)locinput); |
a0ed51b3 |
1478 | } |
1479 | if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8)) |
1480 | sayNO; |
1481 | break; |
bbce6d69 |
1482 | case SPACEL: |
3280af22 |
1483 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1484 | /* FALL THROUGH */ |
a0d0e21e |
1485 | case SPACE: |
3280af22 |
1486 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1487 | sayNO; |
bbce6d69 |
1488 | if (!(OP(scan) == SPACE |
76e3520e |
1489 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
4633a7c4 |
1490 | sayNO; |
76e3520e |
1491 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1492 | break; |
a0ed51b3 |
1493 | case SPACELUTF8: |
1494 | PL_reg_flags |= RF_tainted; |
1495 | /* FALL THROUGH */ |
1496 | case SPACEUTF8: |
1497 | if (!nextchr && locinput >= PL_regeol) |
1498 | sayNO; |
1499 | if (nextchr & 0x80) { |
1500 | if (!(OP(scan) == SPACEUTF8 |
dfe13c55 |
1501 | ? swash_fetch(PL_utf8_space,(U8*)locinput) |
1502 | : isSPACE_LC_utf8((U8*)locinput))) |
1503 | { |
a0ed51b3 |
1504 | sayNO; |
dfe13c55 |
1505 | } |
6f06b55f |
1506 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1507 | nextchr = UCHARAT(locinput); |
1508 | break; |
1509 | } |
1510 | if (!(OP(scan) == SPACEUTF8 |
1511 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
1512 | sayNO; |
1513 | nextchr = UCHARAT(++locinput); |
1514 | break; |
bbce6d69 |
1515 | case NSPACEL: |
3280af22 |
1516 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1517 | /* FALL THROUGH */ |
a0d0e21e |
1518 | case NSPACE: |
76e3520e |
1519 | if (!nextchr) |
4633a7c4 |
1520 | sayNO; |
bbce6d69 |
1521 | if (OP(scan) == SPACE |
76e3520e |
1522 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 |
1523 | sayNO; |
76e3520e |
1524 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1525 | break; |
a0ed51b3 |
1526 | case NSPACELUTF8: |
1527 | PL_reg_flags |= RF_tainted; |
1528 | /* FALL THROUGH */ |
1529 | case NSPACEUTF8: |
1530 | if (!nextchr) |
1531 | sayNO; |
1532 | if (nextchr & 0x80) { |
1533 | if (OP(scan) == NSPACEUTF8 |
dfe13c55 |
1534 | ? swash_fetch(PL_utf8_space,(U8*)locinput) |
1535 | : isSPACE_LC_utf8((U8*)locinput)) |
1536 | { |
a0ed51b3 |
1537 | sayNO; |
dfe13c55 |
1538 | } |
6f06b55f |
1539 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1540 | nextchr = UCHARAT(locinput); |
1541 | break; |
1542 | } |
1543 | if (OP(scan) == NSPACEUTF8 |
1544 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
1545 | sayNO; |
1546 | nextchr = UCHARAT(++locinput); |
1547 | break; |
a0d0e21e |
1548 | case DIGIT: |
76e3520e |
1549 | if (!isDIGIT(nextchr)) |
4633a7c4 |
1550 | sayNO; |
76e3520e |
1551 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1552 | break; |
a0ed51b3 |
1553 | case DIGITUTF8: |
1554 | if (nextchr & 0x80) { |
dfe13c55 |
1555 | if (!(swash_fetch(PL_utf8_digit,(U8*)locinput))) |
a0ed51b3 |
1556 | sayNO; |
6f06b55f |
1557 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1558 | nextchr = UCHARAT(locinput); |
1559 | break; |
1560 | } |
1561 | if (!isDIGIT(nextchr)) |
1562 | sayNO; |
1563 | nextchr = UCHARAT(++locinput); |
1564 | break; |
a0d0e21e |
1565 | case NDIGIT: |
3280af22 |
1566 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1567 | sayNO; |
76e3520e |
1568 | if (isDIGIT(nextchr)) |
4633a7c4 |
1569 | sayNO; |
76e3520e |
1570 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1571 | break; |
a0ed51b3 |
1572 | case NDIGITUTF8: |
1573 | if (!nextchr && locinput >= PL_regeol) |
1574 | sayNO; |
1575 | if (nextchr & 0x80) { |
dfe13c55 |
1576 | if (swash_fetch(PL_utf8_digit,(U8*)locinput)) |
a0ed51b3 |
1577 | sayNO; |
6f06b55f |
1578 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1579 | nextchr = UCHARAT(locinput); |
1580 | break; |
1581 | } |
1582 | if (isDIGIT(nextchr)) |
1583 | sayNO; |
1584 | nextchr = UCHARAT(++locinput); |
1585 | break; |
1586 | case CLUMP: |
dfe13c55 |
1587 | if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput)) |
a0ed51b3 |
1588 | sayNO; |
6f06b55f |
1589 | locinput += PL_utf8skip[nextchr]; |
dfe13c55 |
1590 | while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput)) |
a0ed51b3 |
1591 | locinput += UTF8SKIP(locinput); |
1592 | if (locinput > PL_regeol) |
1593 | sayNO; |
1594 | nextchr = UCHARAT(locinput); |
1595 | break; |
c8756f30 |
1596 | case REFFL: |
3280af22 |
1597 | PL_reg_flags |= RF_tainted; |
c8756f30 |
1598 | /* FALL THROUGH */ |
c277df42 |
1599 | case REF: |
c8756f30 |
1600 | case REFF: |
c277df42 |
1601 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
1602 | ln = PL_regstartp[n]; |
1603 | if (*PL_reglastparen < n || ln == -1) |
af3f8c16 |
1604 | sayNO; /* Do not match unless seen CLOSEn. */ |
cf93c79d |
1605 | if (ln == PL_regendp[n]) |
a0d0e21e |
1606 | break; |
a0ed51b3 |
1607 | |
cf93c79d |
1608 | s = PL_bostr + ln; |
a0ed51b3 |
1609 | if (UTF && OP(scan) != REF) { /* REF can do byte comparison */ |
1610 | char *l = locinput; |
cf93c79d |
1611 | char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 |
1612 | /* |
1613 | * Note that we can't do the "other character" lookup trick as |
1614 | * in the 8-bit case (no pun intended) because in Unicode we |
1615 | * have to map both upper and title case to lower case. |
1616 | */ |
1617 | if (OP(scan) == REFF) { |
1618 | while (s < e) { |
1619 | if (l >= PL_regeol) |
1620 | sayNO; |
dfe13c55 |
1621 | if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l)) |
a0ed51b3 |
1622 | sayNO; |
1623 | s += UTF8SKIP(s); |
1624 | l += UTF8SKIP(l); |
1625 | } |
1626 | } |
1627 | else { |
1628 | while (s < e) { |
1629 | if (l >= PL_regeol) |
1630 | sayNO; |
dfe13c55 |
1631 | if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l)) |
a0ed51b3 |
1632 | sayNO; |
1633 | s += UTF8SKIP(s); |
1634 | l += UTF8SKIP(l); |
1635 | } |
1636 | } |
1637 | locinput = l; |
1638 | nextchr = UCHARAT(locinput); |
1639 | break; |
1640 | } |
1641 | |
a0d0e21e |
1642 | /* Inline the first character, for speed. */ |
76e3520e |
1643 | if (UCHARAT(s) != nextchr && |
c8756f30 |
1644 | (OP(scan) == REF || |
1645 | (UCHARAT(s) != ((OP(scan) == REFF |
22c35a8c |
1646 | ? PL_fold : PL_fold_locale)[nextchr])))) |
4633a7c4 |
1647 | sayNO; |
cf93c79d |
1648 | ln = PL_regendp[n] - ln; |
3280af22 |
1649 | if (locinput + ln > PL_regeol) |
4633a7c4 |
1650 | sayNO; |
c8756f30 |
1651 | if (ln > 1 && (OP(scan) == REF |
1652 | ? memNE(s, locinput, ln) |
1653 | : (OP(scan) == REFF |
1654 | ? ibcmp(s, locinput, ln) |
1655 | : ibcmp_locale(s, locinput, ln)))) |
4633a7c4 |
1656 | sayNO; |
a0d0e21e |
1657 | locinput += ln; |
76e3520e |
1658 | nextchr = UCHARAT(locinput); |
a0d0e21e |
1659 | break; |
1660 | |
1661 | case NOTHING: |
c277df42 |
1662 | case TAIL: |
a0d0e21e |
1663 | break; |
1664 | case BACK: |
1665 | break; |
c277df42 |
1666 | case EVAL: |
1667 | { |
1668 | dSP; |
533c011a |
1669 | OP_4tree *oop = PL_op; |
3280af22 |
1670 | COP *ocurcop = PL_curcop; |
1671 | SV **ocurpad = PL_curpad; |
c277df42 |
1672 | SV *ret; |
1673 | |
1674 | n = ARG(scan); |
533c011a |
1675 | PL_op = (OP_4tree*)PL_regdata->data[n]; |
1676 | DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) ); |
dfad63ad |
1677 | PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]); |
cf93c79d |
1678 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
c277df42 |
1679 | |
cea2e8a9 |
1680 | CALLRUNOPS(aTHX); /* Scalar context. */ |
c277df42 |
1681 | SPAGAIN; |
1682 | ret = POPs; |
1683 | PUTBACK; |
1684 | |
0f5d15d6 |
1685 | PL_op = oop; |
1686 | PL_curpad = ocurpad; |
1687 | PL_curcop = ocurcop; |
c277df42 |
1688 | if (logical) { |
0f5d15d6 |
1689 | if (logical == 2) { /* Postponed subexpression. */ |
1690 | regexp *re; |
22c35a8c |
1691 | MAGIC *mg = Null(MAGIC*); |
0f5d15d6 |
1692 | re_cc_state state; |
1693 | CURCUR cctmp; |
1694 | CHECKPOINT cp, lastcp; |
1695 | |
1696 | if(SvROK(ret) || SvRMAGICAL(ret)) { |
1697 | SV *sv = SvROK(ret) ? SvRV(ret) : ret; |
1698 | |
1699 | if(SvMAGICAL(sv)) |
1700 | mg = mg_find(sv, 'r'); |
1701 | } |
1702 | if (mg) { |
1703 | re = (regexp *)mg->mg_obj; |
df0003d4 |
1704 | (void)ReREFCNT_inc(re); |
0f5d15d6 |
1705 | } |
1706 | else { |
1707 | STRLEN len; |
1708 | char *t = SvPV(ret, len); |
1709 | PMOP pm; |
1710 | char *oprecomp = PL_regprecomp; |
1711 | I32 osize = PL_regsize; |
1712 | I32 onpar = PL_regnpar; |
1713 | |
1714 | pm.op_pmflags = 0; |
cea2e8a9 |
1715 | re = CALLREGCOMP(aTHX_ t, t + len, &pm); |
0f5d15d6 |
1716 | if (!(SvFLAGS(ret) |
1717 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) |
1718 | sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); |
1719 | PL_regprecomp = oprecomp; |
1720 | PL_regsize = osize; |
1721 | PL_regnpar = onpar; |
1722 | } |
1723 | DEBUG_r( |
1724 | PerlIO_printf(Perl_debug_log, |
1725 | "Entering embedded `%s%.60s%s%s'\n", |
1726 | PL_colors[0], |
1727 | re->precomp, |
1728 | PL_colors[1], |
1729 | (strlen(re->precomp) > 60 ? "..." : "")) |
1730 | ); |
1731 | state.node = next; |
1732 | state.prev = PL_reg_call_cc; |
1733 | state.cc = PL_regcc; |
1734 | state.re = PL_reg_re; |
1735 | |
1736 | cctmp.cur = 0; |
1737 | cctmp.oldcc = 0; |
1738 | PL_regcc = &cctmp; |
1739 | |
1740 | cp = regcppush(0); /* Save *all* the positions. */ |
1741 | REGCP_SET; |
1742 | cache_re(re); |
1743 | state.ss = PL_savestack_ix; |
1744 | *PL_reglastparen = 0; |
1745 | PL_reg_call_cc = &state; |
1746 | PL_reginput = locinput; |
1747 | if (regmatch(re->program + 1)) { |
1748 | ReREFCNT_dec(re); |
1749 | regcpblow(cp); |
1750 | sayYES; |
1751 | } |
1752 | DEBUG_r( |
1753 | PerlIO_printf(Perl_debug_log, |
1754 | "%*s failed...\n", |
1755 | REPORT_CODE_OFF+PL_regindent*2, "") |
1756 | ); |
1757 | ReREFCNT_dec(re); |
1758 | REGCP_UNWIND; |
1759 | regcppop(); |
1760 | PL_reg_call_cc = state.prev; |
1761 | PL_regcc = state.cc; |
1762 | PL_reg_re = state.re; |
d3790889 |
1763 | cache_re(PL_reg_re); |
0f5d15d6 |
1764 | sayNO; |
1765 | } |
c277df42 |
1766 | sw = SvTRUE(ret); |
0f5d15d6 |
1767 | logical = 0; |
a0ed51b3 |
1768 | } |
1769 | else |
3280af22 |
1770 | sv_setsv(save_scalar(PL_replgv), ret); |
c277df42 |
1771 | break; |
1772 | } |
a0d0e21e |
1773 | case OPEN: |
c277df42 |
1774 | n = ARG(scan); /* which paren pair */ |
3280af22 |
1775 | PL_reg_start_tmp[n] = locinput; |
1776 | if (n > PL_regsize) |
1777 | PL_regsize = n; |
a0d0e21e |
1778 | break; |
1779 | case CLOSE: |
c277df42 |
1780 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
1781 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; |
1782 | PL_regendp[n] = locinput - PL_bostr; |
3280af22 |
1783 | if (n > *PL_reglastparen) |
1784 | *PL_reglastparen = n; |
a0d0e21e |
1785 | break; |
c277df42 |
1786 | case GROUPP: |
1787 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
1788 | sw = (*PL_reglastparen >= n && PL_regendp[n] != -1); |
c277df42 |
1789 | break; |
1790 | case IFTHEN: |
1791 | if (sw) |
1792 | next = NEXTOPER(NEXTOPER(scan)); |
1793 | else { |
1794 | next = scan + ARG(scan); |
1795 | if (OP(next) == IFTHEN) /* Fake one. */ |
1796 | next = NEXTOPER(NEXTOPER(next)); |
1797 | } |
1798 | break; |
1799 | case LOGICAL: |
0f5d15d6 |
1800 | logical = scan->flags; |
c277df42 |
1801 | break; |
a0d0e21e |
1802 | case CURLYX: { |
1803 | CURCUR cc; |
3280af22 |
1804 | CHECKPOINT cp = PL_savestack_ix; |
c277df42 |
1805 | |
1806 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ |
1807 | next += ARG(next); |
3280af22 |
1808 | cc.oldcc = PL_regcc; |
1809 | PL_regcc = &cc; |
1810 | cc.parenfloor = *PL_reglastparen; |
a0d0e21e |
1811 | cc.cur = -1; |
1812 | cc.min = ARG1(scan); |
1813 | cc.max = ARG2(scan); |
c277df42 |
1814 | cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
a0d0e21e |
1815 | cc.next = next; |
1816 | cc.minmod = minmod; |
1817 | cc.lastloc = 0; |
3280af22 |
1818 | PL_reginput = locinput; |
a0d0e21e |
1819 | n = regmatch(PREVOPER(next)); /* start on the WHILEM */ |
1820 | regcpblow(cp); |
3280af22 |
1821 | PL_regcc = cc.oldcc; |
4633a7c4 |
1822 | saySAME(n); |
a0d0e21e |
1823 | } |
1824 | /* NOT REACHED */ |
1825 | case WHILEM: { |
1826 | /* |
1827 | * This is really hard to understand, because after we match |
1828 | * what we're trying to match, we must make sure the rest of |
1829 | * the RE is going to match for sure, and to do that we have |
1830 | * to go back UP the parse tree by recursing ever deeper. And |
1831 | * if it fails, we have to reset our parent's current state |
1832 | * that we can try again after backing off. |
1833 | */ |
1834 | |
c277df42 |
1835 | CHECKPOINT cp, lastcp; |
3280af22 |
1836 | CURCUR* cc = PL_regcc; |
c277df42 |
1837 | char *lastloc = cc->lastloc; /* Detection of 0-len. */ |
1838 | |
4633a7c4 |
1839 | n = cc->cur + 1; /* how many we know we matched */ |
3280af22 |
1840 | PL_reginput = locinput; |
a0d0e21e |
1841 | |
c277df42 |
1842 | DEBUG_r( |
1843 | PerlIO_printf(Perl_debug_log, |
1844 | "%*s %ld out of %ld..%ld cc=%lx\n", |
3280af22 |
1845 | REPORT_CODE_OFF+PL_regindent*2, "", |
c277df42 |
1846 | (long)n, (long)cc->min, |
1847 | (long)cc->max, (long)cc) |
1848 | ); |
4633a7c4 |
1849 | |
a0d0e21e |
1850 | /* If degenerate scan matches "", assume scan done. */ |
1851 | |
579cf2c3 |
1852 | if (locinput == cc->lastloc && n >= cc->min) { |
3280af22 |
1853 | PL_regcc = cc->oldcc; |
1854 | ln = PL_regcc->cur; |
c277df42 |
1855 | DEBUG_r( |
c3464db5 |
1856 | PerlIO_printf(Perl_debug_log, |
1857 | "%*s empty match detected, try continuation...\n", |
3280af22 |
1858 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1859 | ); |
a0d0e21e |
1860 | if (regmatch(cc->next)) |
4633a7c4 |
1861 | sayYES; |
c277df42 |
1862 | DEBUG_r( |
c3464db5 |
1863 | PerlIO_printf(Perl_debug_log, |
1864 | "%*s failed...\n", |
3280af22 |
1865 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1866 | ); |
3280af22 |
1867 | PL_regcc->cur = ln; |
1868 | PL_regcc = cc; |
4633a7c4 |
1869 | sayNO; |
a0d0e21e |
1870 | } |
1871 | |
1872 | /* First just match a string of min scans. */ |
1873 | |
1874 | if (n < cc->min) { |
1875 | cc->cur = n; |
1876 | cc->lastloc = locinput; |
4633a7c4 |
1877 | if (regmatch(cc->scan)) |
1878 | sayYES; |
1879 | cc->cur = n - 1; |
c277df42 |
1880 | cc->lastloc = lastloc; |
1881 | DEBUG_r( |
c3464db5 |
1882 | PerlIO_printf(Perl_debug_log, |
1883 | "%*s failed...\n", |
3280af22 |
1884 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1885 | ); |
4633a7c4 |
1886 | sayNO; |
a0d0e21e |
1887 | } |
1888 | |
1889 | /* Prefer next over scan for minimal matching. */ |
1890 | |
1891 | if (cc->minmod) { |
3280af22 |
1892 | PL_regcc = cc->oldcc; |
1893 | ln = PL_regcc->cur; |
5f05dabc |
1894 | cp = regcppush(cc->parenfloor); |
c277df42 |
1895 | REGCP_SET; |
5f05dabc |
1896 | if (regmatch(cc->next)) { |
c277df42 |
1897 | regcpblow(cp); |
4633a7c4 |
1898 | sayYES; /* All done. */ |
5f05dabc |
1899 | } |
c277df42 |
1900 | REGCP_UNWIND; |
5f05dabc |
1901 | regcppop(); |
3280af22 |
1902 | PL_regcc->cur = ln; |
1903 | PL_regcc = cc; |
a0d0e21e |
1904 | |
c277df42 |
1905 | if (n >= cc->max) { /* Maximum greed exceeded? */ |
599cee73 |
1906 | if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY |
3280af22 |
1907 | && !(PL_reg_flags & RF_warned)) { |
1908 | PL_reg_flags |= RF_warned; |
cea2e8a9 |
1909 | Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", |
2f3ca594 |
1910 | "Complex regular subexpression recursion", |
1911 | REG_INFTY - 1); |
c277df42 |
1912 | } |
4633a7c4 |
1913 | sayNO; |
c277df42 |
1914 | } |
a687059c |
1915 | |
c277df42 |
1916 | DEBUG_r( |
c3464db5 |
1917 | PerlIO_printf(Perl_debug_log, |
1918 | "%*s trying longer...\n", |
3280af22 |
1919 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1920 | ); |
a0d0e21e |
1921 | /* Try scanning more and see if it helps. */ |
3280af22 |
1922 | PL_reginput = locinput; |
a0d0e21e |
1923 | cc->cur = n; |
1924 | cc->lastloc = locinput; |
5f05dabc |
1925 | cp = regcppush(cc->parenfloor); |
c277df42 |
1926 | REGCP_SET; |
5f05dabc |
1927 | if (regmatch(cc->scan)) { |
c277df42 |
1928 | regcpblow(cp); |
4633a7c4 |
1929 | sayYES; |
5f05dabc |
1930 | } |
c277df42 |
1931 | DEBUG_r( |
c3464db5 |
1932 | PerlIO_printf(Perl_debug_log, |
1933 | "%*s failed...\n", |
3280af22 |
1934 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1935 | ); |
1936 | REGCP_UNWIND; |
5f05dabc |
1937 | regcppop(); |
4633a7c4 |
1938 | cc->cur = n - 1; |
c277df42 |
1939 | cc->lastloc = lastloc; |
4633a7c4 |
1940 | sayNO; |
a0d0e21e |
1941 | } |
1942 | |
1943 | /* Prefer scan over next for maximal matching. */ |
1944 | |
1945 | if (n < cc->max) { /* More greed allowed? */ |
5f05dabc |
1946 | cp = regcppush(cc->parenfloor); |
a0d0e21e |
1947 | cc->cur = n; |
1948 | cc->lastloc = locinput; |
c277df42 |
1949 | REGCP_SET; |
5f05dabc |
1950 | if (regmatch(cc->scan)) { |
c277df42 |
1951 | regcpblow(cp); |
4633a7c4 |
1952 | sayYES; |
5f05dabc |
1953 | } |
c277df42 |
1954 | REGCP_UNWIND; |
a0d0e21e |
1955 | regcppop(); /* Restore some previous $<digit>s? */ |
3280af22 |
1956 | PL_reginput = locinput; |
c277df42 |
1957 | DEBUG_r( |
c3464db5 |
1958 | PerlIO_printf(Perl_debug_log, |
1959 | "%*s failed, try continuation...\n", |
3280af22 |
1960 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1961 | ); |
1962 | } |
599cee73 |
1963 | if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY |
1964 | && !(PL_reg_flags & RF_warned)) { |
3280af22 |
1965 | PL_reg_flags |= RF_warned; |
cea2e8a9 |
1966 | Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", |
cb5d145d |
1967 | "Complex regular subexpression recursion", |
1968 | REG_INFTY - 1); |
a0d0e21e |
1969 | } |
1970 | |
1971 | /* Failed deeper matches of scan, so see if this one works. */ |
3280af22 |
1972 | PL_regcc = cc->oldcc; |
1973 | ln = PL_regcc->cur; |
a0d0e21e |
1974 | if (regmatch(cc->next)) |
4633a7c4 |
1975 | sayYES; |
c277df42 |
1976 | DEBUG_r( |
c3464db5 |
1977 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", |
3280af22 |
1978 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1979 | ); |
3280af22 |
1980 | PL_regcc->cur = ln; |
1981 | PL_regcc = cc; |
4633a7c4 |
1982 | cc->cur = n - 1; |
c277df42 |
1983 | cc->lastloc = lastloc; |
4633a7c4 |
1984 | sayNO; |
a0d0e21e |
1985 | } |
1986 | /* NOT REACHED */ |
c277df42 |
1987 | case BRANCHJ: |
1988 | next = scan + ARG(scan); |
1989 | if (next == scan) |
1990 | next = NULL; |
1991 | inner = NEXTOPER(NEXTOPER(scan)); |
1992 | goto do_branch; |
1993 | case BRANCH: |
1994 | inner = NEXTOPER(scan); |
1995 | do_branch: |
1996 | { |
1997 | CHECKPOINT lastcp; |
1998 | c1 = OP(scan); |
1999 | if (OP(next) != c1) /* No choice. */ |
2000 | next = inner; /* Avoid recursion. */ |
a0d0e21e |
2001 | else { |
3280af22 |
2002 | int lastparen = *PL_reglastparen; |
c277df42 |
2003 | |
2004 | REGCP_SET; |
a0d0e21e |
2005 | do { |
3280af22 |
2006 | PL_reginput = locinput; |
c277df42 |
2007 | if (regmatch(inner)) |
4633a7c4 |
2008 | sayYES; |
c277df42 |
2009 | REGCP_UNWIND; |
3280af22 |
2010 | for (n = *PL_reglastparen; n > lastparen; n--) |
cf93c79d |
2011 | PL_regendp[n] = -1; |
3280af22 |
2012 | *PL_reglastparen = n; |
c277df42 |
2013 | scan = next; |
a0d0e21e |
2014 | /*SUPPRESS 560*/ |
c277df42 |
2015 | if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))) |
2016 | next += n; |
a0d0e21e |
2017 | else |
c277df42 |
2018 | next = NULL; |
c277df42 |
2019 | inner = NEXTOPER(scan); |
2020 | if (c1 == BRANCHJ) { |
2021 | inner = NEXTOPER(inner); |
2022 | } |
2023 | } while (scan != NULL && OP(scan) == c1); |
4633a7c4 |
2024 | sayNO; |
a0d0e21e |
2025 | /* NOTREACHED */ |
a687059c |
2026 | } |
a0d0e21e |
2027 | } |
2028 | break; |
2029 | case MINMOD: |
2030 | minmod = 1; |
2031 | break; |
c277df42 |
2032 | case CURLYM: |
2033 | { |
00db4c45 |
2034 | I32 l = 0; |
c277df42 |
2035 | CHECKPOINT lastcp; |
2036 | |
2037 | /* We suppose that the next guy does not need |
2038 | backtracking: in particular, it is of constant length, |
2039 | and has no parenths to influence future backrefs. */ |
2040 | ln = ARG1(scan); /* min to match */ |
2041 | n = ARG2(scan); /* max to match */ |
c277df42 |
2042 | paren = scan->flags; |
2043 | if (paren) { |
3280af22 |
2044 | if (paren > PL_regsize) |
2045 | PL_regsize = paren; |
2046 | if (paren > *PL_reglastparen) |
2047 | *PL_reglastparen = paren; |
c277df42 |
2048 | } |
dc45a647 |
2049 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
c277df42 |
2050 | if (paren) |
2051 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
3280af22 |
2052 | PL_reginput = locinput; |
c277df42 |
2053 | if (minmod) { |
2054 | minmod = 0; |
2055 | if (ln && regrepeat_hard(scan, ln, &l) < ln) |
2056 | sayNO; |
5f4b28b2 |
2057 | if (ln && l == 0 && n >= ln |
c277df42 |
2058 | /* In fact, this is tricky. If paren, then the |
2059 | fact that we did/didnot match may influence |
2060 | future execution. */ |
2061 | && !(paren && ln == 0)) |
2062 | ln = n; |
3280af22 |
2063 | locinput = PL_reginput; |
22c35a8c |
2064 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
c277df42 |
2065 | c1 = UCHARAT(OPERAND(next) + 1); |
2066 | if (OP(next) == EXACTF) |
22c35a8c |
2067 | c2 = PL_fold[c1]; |
c277df42 |
2068 | else if (OP(next) == EXACTFL) |
22c35a8c |
2069 | c2 = PL_fold_locale[c1]; |
c277df42 |
2070 | else |
2071 | c2 = c1; |
a0ed51b3 |
2072 | } |
2073 | else |
c277df42 |
2074 | c1 = c2 = -1000; |
2075 | REGCP_SET; |
5f4b28b2 |
2076 | /* This may be improved if l == 0. */ |
c277df42 |
2077 | while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */ |
2078 | /* If it could work, try it. */ |
2079 | if (c1 == -1000 || |
3280af22 |
2080 | UCHARAT(PL_reginput) == c1 || |
2081 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2082 | { |
2083 | if (paren) { |
2084 | if (n) { |
cf93c79d |
2085 | PL_regstartp[paren] = |
2086 | HOPc(PL_reginput, -l) - PL_bostr; |
2087 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2088 | } |
2089 | else |
cf93c79d |
2090 | PL_regendp[paren] = -1; |
c277df42 |
2091 | } |
2092 | if (regmatch(next)) |
2093 | sayYES; |
2094 | REGCP_UNWIND; |
2095 | } |
2096 | /* Couldn't or didn't -- move forward. */ |
3280af22 |
2097 | PL_reginput = locinput; |
c277df42 |
2098 | if (regrepeat_hard(scan, 1, &l)) { |
2099 | ln++; |
3280af22 |
2100 | locinput = PL_reginput; |
c277df42 |
2101 | } |
2102 | else |
2103 | sayNO; |
2104 | } |
a0ed51b3 |
2105 | } |
2106 | else { |
c277df42 |
2107 | n = regrepeat_hard(scan, n, &l); |
2108 | if (n != 0 && l == 0 |
2109 | /* In fact, this is tricky. If paren, then the |
2110 | fact that we did/didnot match may influence |
2111 | future execution. */ |
2112 | && !(paren && ln == 0)) |
2113 | ln = n; |
3280af22 |
2114 | locinput = PL_reginput; |
c277df42 |
2115 | DEBUG_r( |
5c0ca799 |
2116 | PerlIO_printf(Perl_debug_log, |
2117 | "%*s matched %ld times, len=%ld...\n", |
3280af22 |
2118 | REPORT_CODE_OFF+PL_regindent*2, "", n, l) |
c277df42 |
2119 | ); |
2120 | if (n >= ln) { |
22c35a8c |
2121 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
c277df42 |
2122 | c1 = UCHARAT(OPERAND(next) + 1); |
2123 | if (OP(next) == EXACTF) |
22c35a8c |
2124 | c2 = PL_fold[c1]; |
c277df42 |
2125 | else if (OP(next) == EXACTFL) |
22c35a8c |
2126 | c2 = PL_fold_locale[c1]; |
c277df42 |
2127 | else |
2128 | c2 = c1; |
a0ed51b3 |
2129 | } |
2130 | else |
c277df42 |
2131 | c1 = c2 = -1000; |
2132 | } |
2133 | REGCP_SET; |
2134 | while (n >= ln) { |
2135 | /* If it could work, try it. */ |
2136 | if (c1 == -1000 || |
3280af22 |
2137 | UCHARAT(PL_reginput) == c1 || |
2138 | UCHARAT(PL_reginput) == c2) |
a0ed51b3 |
2139 | { |
2140 | DEBUG_r( |
c3464db5 |
2141 | PerlIO_printf(Perl_debug_log, |
2142 | "%*s trying tail with n=%ld...\n", |
3280af22 |
2143 | REPORT_CODE_OFF+PL_regindent*2, "", n) |
a0ed51b3 |
2144 | ); |
2145 | if (paren) { |
2146 | if (n) { |
cf93c79d |
2147 | PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr; |
2148 | PL_regendp[paren] = PL_reginput - PL_bostr; |
c277df42 |
2149 | } |
a0ed51b3 |
2150 | else |
cf93c79d |
2151 | PL_regendp[paren] = -1; |
c277df42 |
2152 | } |
a0ed51b3 |
2153 | if (regmatch(next)) |
2154 | sayYES; |
2155 | REGCP_UNWIND; |
2156 | } |
c277df42 |
2157 | /* Couldn't or didn't -- back up. */ |
2158 | n--; |
dfe13c55 |
2159 | locinput = HOPc(locinput, -l); |
3280af22 |
2160 | PL_reginput = locinput; |
c277df42 |
2161 | } |
2162 | } |
2163 | sayNO; |
2164 | break; |
2165 | } |
2166 | case CURLYN: |
2167 | paren = scan->flags; /* Which paren to set */ |
3280af22 |
2168 | if (paren > PL_regsize) |
2169 | PL_regsize = paren; |
2170 | if (paren > *PL_reglastparen) |
2171 | *PL_reglastparen = paren; |
c277df42 |
2172 | ln = ARG1(scan); /* min to match */ |
2173 | n = ARG2(scan); /* max to match */ |
dc45a647 |
2174 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
c277df42 |
2175 | goto repeat; |
a0d0e21e |
2176 | case CURLY: |
c277df42 |
2177 | paren = 0; |
a0d0e21e |
2178 | ln = ARG1(scan); /* min to match */ |
2179 | n = ARG2(scan); /* max to match */ |
dc45a647 |
2180 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
a0d0e21e |
2181 | goto repeat; |
2182 | case STAR: |
2183 | ln = 0; |
c277df42 |
2184 | n = REG_INFTY; |
a0d0e21e |
2185 | scan = NEXTOPER(scan); |
c277df42 |
2186 | paren = 0; |
a0d0e21e |
2187 | goto repeat; |
2188 | case PLUS: |
c277df42 |
2189 | ln = 1; |
2190 | n = REG_INFTY; |
2191 | scan = NEXTOPER(scan); |
2192 | paren = 0; |
2193 | repeat: |
a0d0e21e |
2194 | /* |
2195 | * Lookahead to avoid useless match attempts |
2196 | * when we know what character comes next. |
2197 | */ |
22c35a8c |
2198 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
bbce6d69 |
2199 | c1 = UCHARAT(OPERAND(next) + 1); |
2200 | if (OP(next) == EXACTF) |
22c35a8c |
2201 | c2 = PL_fold[c1]; |
bbce6d69 |
2202 | else if (OP(next) == EXACTFL) |
22c35a8c |
2203 | c2 = PL_fold_locale[c1]; |
bbce6d69 |
2204 | else |
2205 | c2 = c1; |
2206 | } |
a0d0e21e |
2207 | else |
bbce6d69 |
2208 | c1 = c2 = -1000; |
3280af22 |
2209 | PL_reginput = locinput; |
a0d0e21e |
2210 | if (minmod) { |
c277df42 |
2211 | CHECKPOINT lastcp; |
a0d0e21e |
2212 | minmod = 0; |
2213 | if (ln && regrepeat(scan, ln) < ln) |
4633a7c4 |
2214 | sayNO; |
a0ed51b3 |
2215 | locinput = PL_reginput; |
c277df42 |
2216 | REGCP_SET; |
0fe9bf95 |
2217 | if (c1 != -1000) { |
2218 | char *e = locinput + n - ln; /* Should not check after this */ |
2219 | char *old = locinput; |
2220 | |
2221 | if (e >= PL_regeol || (n == REG_INFTY)) |
2222 | e = PL_regeol - 1; |
2223 | while (1) { |
2224 | /* Find place 'next' could work */ |
2225 | if (c1 == c2) { |
2226 | while (locinput <= e && *locinput != c1) |
2227 | locinput++; |
2228 | } else { |
2229 | while (locinput <= e |
2230 | && *locinput != c1 |
2231 | && *locinput != c2) |
2232 | locinput++; |
2233 | } |
2234 | if (locinput > e) |
2235 | sayNO; |
2236 | /* PL_reginput == old now */ |
2237 | if (locinput != old) { |
2238 | ln = 1; /* Did some */ |
2239 | if (regrepeat(scan, locinput - old) < |
2240 | locinput - old) |
2241 | sayNO; |
2242 | } |
2243 | /* PL_reginput == locinput now */ |
2244 | if (paren) { |
2245 | if (ln) { |
cf93c79d |
2246 | PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; |
2247 | PL_regendp[paren] = locinput - PL_bostr; |
0fe9bf95 |
2248 | } |
2249 | else |
cf93c79d |
2250 | PL_regendp[paren] = -1; |
0fe9bf95 |
2251 | } |
2252 | if (regmatch(next)) |
2253 | sayYES; |
2254 | PL_reginput = locinput; /* Could be reset... */ |
2255 | REGCP_UNWIND; |
2256 | /* Couldn't or didn't -- move forward. */ |
2257 | old = locinput++; |
2258 | } |
2259 | } |
2260 | else |
c277df42 |
2261 | while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ |
a0d0e21e |
2262 | /* If it could work, try it. */ |
bbce6d69 |
2263 | if (c1 == -1000 || |
3280af22 |
2264 | UCHARAT(PL_reginput) == c1 || |
2265 | UCHARAT(PL_reginput) == c2) |
bbce6d69 |
2266 | { |
c277df42 |
2267 | if (paren) { |
2268 | if (n) { |
cf93c79d |
2269 | PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; |
2270 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2271 | } |
2272 | else |
cf93c79d |
2273 | PL_regendp[paren] = -1; |
c277df42 |
2274 | } |
a0d0e21e |
2275 | if (regmatch(next)) |
4633a7c4 |
2276 | sayYES; |
c277df42 |
2277 | REGCP_UNWIND; |
bbce6d69 |
2278 | } |
c277df42 |
2279 | /* Couldn't or didn't -- move forward. */ |
a0ed51b3 |
2280 | PL_reginput = locinput; |
a0d0e21e |
2281 | if (regrepeat(scan, 1)) { |
2282 | ln++; |
a0ed51b3 |
2283 | locinput = PL_reginput; |
2284 | } |
2285 | else |
4633a7c4 |
2286 | sayNO; |
a0d0e21e |
2287 | } |
2288 | } |
2289 | else { |
c277df42 |
2290 | CHECKPOINT lastcp; |
a0d0e21e |
2291 | n = regrepeat(scan, n); |
a0ed51b3 |
2292 | locinput = PL_reginput; |
22c35a8c |
2293 | if (ln < n && PL_regkind[(U8)OP(next)] == EOL && |
3280af22 |
2294 | (!PL_multiline || OP(next) == SEOL)) |
a0d0e21e |
2295 | ln = n; /* why back off? */ |
c277df42 |
2296 | REGCP_SET; |
2297 | if (paren) { |
2298 | while (n >= ln) { |
2299 | /* If it could work, try it. */ |
2300 | if (c1 == -1000 || |
3280af22 |
2301 | UCHARAT(PL_reginput) == c1 || |
2302 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2303 | { |
2304 | if (paren && n) { |
2305 | if (n) { |
cf93c79d |
2306 | PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; |
2307 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2308 | } |
2309 | else |
cf93c79d |
2310 | PL_regendp[paren] = -1; |
c277df42 |
2311 | } |
2312 | if (regmatch(next)) |
2313 | sayYES; |
2314 | REGCP_UNWIND; |
2315 | } |
2316 | /* Couldn't or didn't -- back up. */ |
2317 | n--; |
dfe13c55 |
2318 | PL_reginput = locinput = HOPc(locinput, -1); |
c277df42 |
2319 | } |
a0ed51b3 |
2320 | } |
2321 | else { |
c277df42 |
2322 | while (n >= ln) { |
2323 | /* If it could work, try it. */ |
2324 | if (c1 == -1000 || |
3280af22 |
2325 | UCHARAT(PL_reginput) == c1 || |
2326 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2327 | { |
2328 | if (regmatch(next)) |
2329 | sayYES; |
2330 | REGCP_UNWIND; |
2331 | } |
2332 | /* Couldn't or didn't -- back up. */ |
2333 | n--; |
dfe13c55 |
2334 | PL_reginput = locinput = HOPc(locinput, -1); |
bbce6d69 |
2335 | } |
a0d0e21e |
2336 | } |
2337 | } |
4633a7c4 |
2338 | sayNO; |
c277df42 |
2339 | break; |
a0d0e21e |
2340 | case END: |
0f5d15d6 |
2341 | if (PL_reg_call_cc) { |
2342 | re_cc_state *cur_call_cc = PL_reg_call_cc; |
2343 | CURCUR *cctmp = PL_regcc; |
2344 | regexp *re = PL_reg_re; |
2345 | CHECKPOINT cp, lastcp; |
2346 | |
2347 | cp = regcppush(0); /* Save *all* the positions. */ |
2348 | REGCP_SET; |
2349 | regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of |
2350 | the caller. */ |
2351 | PL_reginput = locinput; /* Make position available to |
2352 | the callcc. */ |
2353 | cache_re(PL_reg_call_cc->re); |
2354 | PL_regcc = PL_reg_call_cc->cc; |
2355 | PL_reg_call_cc = PL_reg_call_cc->prev; |
2356 | if (regmatch(cur_call_cc->node)) { |
2357 | PL_reg_call_cc = cur_call_cc; |
2358 | regcpblow(cp); |
2359 | sayYES; |
2360 | } |
2361 | REGCP_UNWIND; |
2362 | regcppop(); |
2363 | PL_reg_call_cc = cur_call_cc; |
2364 | PL_regcc = cctmp; |
2365 | PL_reg_re = re; |
2366 | cache_re(re); |
2367 | |
2368 | DEBUG_r( |
2369 | PerlIO_printf(Perl_debug_log, |
2370 | "%*s continuation failed...\n", |
2371 | REPORT_CODE_OFF+PL_regindent*2, "") |
2372 | ); |
2373 | sayNO; |
2374 | } |
3280af22 |
2375 | if (locinput < PL_regtill) |
7e5428c5 |
2376 | sayNO; /* Cannot match: too short. */ |
2377 | /* Fall through */ |
2378 | case SUCCEED: |
3280af22 |
2379 | PL_reginput = locinput; /* put where regtry can find it */ |
4633a7c4 |
2380 | sayYES; /* Success! */ |
c277df42 |
2381 | case SUSPEND: |
2382 | n = 1; |
9fe1d20c |
2383 | PL_reginput = locinput; |
c277df42 |
2384 | goto do_ifmatch; |
a0d0e21e |
2385 | case UNLESSM: |
c277df42 |
2386 | n = 0; |
a0ed51b3 |
2387 | if (scan->flags) { |
0fe9bf95 |
2388 | if (UTF) { /* XXXX This is absolutely |
2389 | broken, we read before |
2390 | start of string. */ |
2391 | s = HOPMAYBEc(locinput, -scan->flags); |
2392 | if (!s) |
2393 | goto say_yes; |
2394 | PL_reginput = s; |
2395 | } |
2396 | else { |
2397 | if (locinput < PL_bostr + scan->flags) |
2398 | goto say_yes; |
2399 | PL_reginput = locinput - scan->flags; |
2400 | goto do_ifmatch; |
2401 | } |
a0ed51b3 |
2402 | } |
2403 | else |
2404 | PL_reginput = locinput; |
c277df42 |
2405 | goto do_ifmatch; |
2406 | case IFMATCH: |
2407 | n = 1; |
a0ed51b3 |
2408 | if (scan->flags) { |
0fe9bf95 |
2409 | if (UTF) { /* XXXX This is absolutely |
2410 | broken, we read before |
2411 | start of string. */ |
2412 | s = HOPMAYBEc(locinput, -scan->flags); |
2413 | if (!s || s < PL_bostr) |
2414 | goto say_no; |
2415 | PL_reginput = s; |
2416 | } |
2417 | else { |
2418 | if (locinput < PL_bostr + scan->flags) |
2419 | goto say_no; |
2420 | PL_reginput = locinput - scan->flags; |
2421 | goto do_ifmatch; |
2422 | } |
a0ed51b3 |
2423 | } |
2424 | else |
2425 | PL_reginput = locinput; |
2426 | |
c277df42 |
2427 | do_ifmatch: |
c277df42 |
2428 | inner = NEXTOPER(NEXTOPER(scan)); |
2429 | if (regmatch(inner) != n) { |
2430 | say_no: |
2431 | if (logical) { |
2432 | logical = 0; |
2433 | sw = 0; |
2434 | goto do_longjump; |
a0ed51b3 |
2435 | } |
2436 | else |
c277df42 |
2437 | sayNO; |
2438 | } |
2439 | say_yes: |
2440 | if (logical) { |
2441 | logical = 0; |
2442 | sw = 1; |
2443 | } |
fe44a5e8 |
2444 | if (OP(scan) == SUSPEND) { |
3280af22 |
2445 | locinput = PL_reginput; |
565764a8 |
2446 | nextchr = UCHARAT(locinput); |
fe44a5e8 |
2447 | } |
c277df42 |
2448 | /* FALL THROUGH. */ |
2449 | case LONGJMP: |
2450 | do_longjump: |
2451 | next = scan + ARG(scan); |
2452 | if (next == scan) |
2453 | next = NULL; |
a0d0e21e |
2454 | break; |
2455 | default: |
c030ccd9 |
2456 | PerlIO_printf(PerlIO_stderr(), "%lx %d\n", |
c277df42 |
2457 | (unsigned long)scan, OP(scan)); |
cea2e8a9 |
2458 | Perl_croak(aTHX_ "regexp memory corruption"); |
a687059c |
2459 | } |
a0d0e21e |
2460 | scan = next; |
2461 | } |
a687059c |
2462 | |
a0d0e21e |
2463 | /* |
2464 | * We get here only if there's trouble -- normally "case END" is |
2465 | * the terminating point. |
2466 | */ |
cea2e8a9 |
2467 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e |
2468 | /*NOTREACHED*/ |
4633a7c4 |
2469 | sayNO; |
2470 | |
2471 | yes: |
2472 | #ifdef DEBUGGING |
3280af22 |
2473 | PL_regindent--; |
4633a7c4 |
2474 | #endif |
2475 | return 1; |
2476 | |
2477 | no: |
2478 | #ifdef DEBUGGING |
3280af22 |
2479 | PL_regindent--; |
4633a7c4 |
2480 | #endif |
a0d0e21e |
2481 | return 0; |
a687059c |
2482 | } |
2483 | |
2484 | /* |
2485 | - regrepeat - repeatedly match something simple, report how many |
2486 | */ |
2487 | /* |
2488 | * [This routine now assumes that it will only match on things of length 1. |
2489 | * That was true before, but now we assume scan - reginput is the count, |
a0ed51b3 |
2490 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c |
2491 | */ |
76e3520e |
2492 | STATIC I32 |
cea2e8a9 |
2493 | S_regrepeat(pTHX_ regnode *p, I32 max) |
a687059c |
2494 | { |
5c0ca799 |
2495 | dTHR; |
a0d0e21e |
2496 | register char *scan; |
2497 | register char *opnd; |
2498 | register I32 c; |
3280af22 |
2499 | register char *loceol = PL_regeol; |
a0ed51b3 |
2500 | register I32 hardcount = 0; |
a0d0e21e |
2501 | |
3280af22 |
2502 | scan = PL_reginput; |
c277df42 |
2503 | if (max != REG_INFTY && max < loceol - scan) |
a0d0e21e |
2504 | loceol = scan + max; |
161b471a |
2505 | opnd = (char *) OPERAND(p); |
a0d0e21e |
2506 | switch (OP(p)) { |
22c35a8c |
2507 | case REG_ANY: |
a0d0e21e |
2508 | while (scan < loceol && *scan != '\n') |
2509 | scan++; |
2510 | break; |
2511 | case SANY: |
2512 | scan = loceol; |
2513 | break; |
a0ed51b3 |
2514 | case ANYUTF8: |
2515 | loceol = PL_regeol; |
2516 | while (scan < loceol && *scan != '\n') { |
2517 | scan += UTF8SKIP(scan); |
2518 | hardcount++; |
2519 | } |
2520 | break; |
2521 | case SANYUTF8: |
2522 | loceol = PL_regeol; |
2523 | while (scan < loceol) { |
2524 | scan += UTF8SKIP(scan); |
2525 | hardcount++; |
2526 | } |
2527 | break; |
bbce6d69 |
2528 | case EXACT: /* length of string is 1 */ |
2529 | c = UCHARAT(++opnd); |
2530 | while (scan < loceol && UCHARAT(scan) == c) |
2531 | scan++; |
2532 | break; |
2533 | case EXACTF: /* length of string is 1 */ |
2534 | c = UCHARAT(++opnd); |
2535 | while (scan < loceol && |
22c35a8c |
2536 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 |
2537 | scan++; |
2538 | break; |
2539 | case EXACTFL: /* length of string is 1 */ |
3280af22 |
2540 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2541 | c = UCHARAT(++opnd); |
2542 | while (scan < loceol && |
22c35a8c |
2543 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e |
2544 | scan++; |
2545 | break; |
a0ed51b3 |
2546 | case ANYOFUTF8: |
2547 | loceol = PL_regeol; |
2548 | while (scan < loceol && REGINCLASSUTF8(p, (U8*)scan)) { |
2549 | scan += UTF8SKIP(scan); |
2550 | hardcount++; |
2551 | } |
2552 | break; |
a0d0e21e |
2553 | case ANYOF: |
ae5c130c |
2554 | while (scan < loceol && REGINCLASS(opnd, *scan)) |
a0d0e21e |
2555 | scan++; |
a0d0e21e |
2556 | break; |
2557 | case ALNUM: |
2558 | while (scan < loceol && isALNUM(*scan)) |
2559 | scan++; |
2560 | break; |
a0ed51b3 |
2561 | case ALNUMUTF8: |
2562 | loceol = PL_regeol; |
dfe13c55 |
2563 | while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) { |
a0ed51b3 |
2564 | scan += UTF8SKIP(scan); |
2565 | hardcount++; |
2566 | } |
2567 | break; |
bbce6d69 |
2568 | case ALNUML: |
3280af22 |
2569 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2570 | while (scan < loceol && isALNUM_LC(*scan)) |
2571 | scan++; |
2572 | break; |
a0ed51b3 |
2573 | case ALNUMLUTF8: |
2574 | PL_reg_flags |= RF_tainted; |
2575 | loceol = PL_regeol; |
dfe13c55 |
2576 | while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) { |
a0ed51b3 |
2577 | scan += UTF8SKIP(scan); |
2578 | hardcount++; |
2579 | } |
2580 | break; |
2581 | break; |
a0d0e21e |
2582 | case NALNUM: |
2583 | while (scan < loceol && !isALNUM(*scan)) |
2584 | scan++; |
2585 | break; |
a0ed51b3 |
2586 | case NALNUMUTF8: |
2587 | loceol = PL_regeol; |
dfe13c55 |
2588 | while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) { |
a0ed51b3 |
2589 | scan += UTF8SKIP(scan); |
2590 | hardcount++; |
2591 | } |
2592 | break; |
bbce6d69 |
2593 | case NALNUML: |
3280af22 |
2594 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2595 | while (scan < loceol && !isALNUM_LC(*scan)) |
2596 | scan++; |
2597 | break; |
a0ed51b3 |
2598 | case NALNUMLUTF8: |
2599 | PL_reg_flags |= RF_tainted; |
2600 | loceol = PL_regeol; |
dfe13c55 |
2601 | while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) { |
a0ed51b3 |
2602 | scan += UTF8SKIP(scan); |
2603 | hardcount++; |
2604 | } |
2605 | break; |
a0d0e21e |
2606 | case SPACE: |
2607 | while (scan < loceol && isSPACE(*scan)) |
2608 | scan++; |
2609 | break; |
a0ed51b3 |
2610 | case SPACEUTF8: |
2611 | loceol = PL_regeol; |
dfe13c55 |
2612 | while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { |
a0ed51b3 |
2613 | scan += UTF8SKIP(scan); |
2614 | hardcount++; |
2615 | } |
2616 | break; |
bbce6d69 |
2617 | case SPACEL: |
3280af22 |
2618 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2619 | while (scan < loceol && isSPACE_LC(*scan)) |
2620 | scan++; |
2621 | break; |
a0ed51b3 |
2622 | case SPACELUTF8: |
2623 | PL_reg_flags |= RF_tainted; |
2624 | loceol = PL_regeol; |
dfe13c55 |
2625 | while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
a0ed51b3 |
2626 | scan += UTF8SKIP(scan); |
2627 | hardcount++; |
2628 | } |
2629 | break; |
a0d0e21e |
2630 | case NSPACE: |
2631 | while (scan < loceol && !isSPACE(*scan)) |
2632 | scan++; |
2633 | break; |
a0ed51b3 |
2634 | case NSPACEUTF8: |
2635 | loceol = PL_regeol; |
dfe13c55 |
2636 | while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { |
a0ed51b3 |
2637 | scan += UTF8SKIP(scan); |
2638 | hardcount++; |
2639 | } |
2640 | break; |
bbce6d69 |
2641 | case NSPACEL: |
3280af22 |
2642 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2643 | while (scan < loceol && !isSPACE_LC(*scan)) |
2644 | scan++; |
2645 | break; |
a0ed51b3 |
2646 | case NSPACELUTF8: |
2647 | PL_reg_flags |= RF_tainted; |
2648 | loceol = PL_regeol; |
dfe13c55 |
2649 | while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
a0ed51b3 |
2650 | scan += UTF8SKIP(scan); |
2651 | hardcount++; |
2652 | } |
2653 | break; |
a0d0e21e |
2654 | case DIGIT: |
2655 | while (scan < loceol && isDIGIT(*scan)) |
2656 | scan++; |
2657 | break; |
a0ed51b3 |
2658 | case DIGITUTF8: |
2659 | loceol = PL_regeol; |
dfe13c55 |
2660 | while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) { |
a0ed51b3 |
2661 | scan += UTF8SKIP(scan); |
2662 | hardcount++; |
2663 | } |
2664 | break; |
2665 | break; |
a0d0e21e |
2666 | case NDIGIT: |
2667 | while (scan < loceol && !isDIGIT(*scan)) |
2668 | scan++; |
2669 | break; |
a0ed51b3 |
2670 | case NDIGITUTF8: |
2671 | loceol = PL_regeol; |
dfe13c55 |
2672 | while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) { |
a0ed51b3 |
2673 | scan += UTF8SKIP(scan); |
2674 | hardcount++; |
2675 | } |
2676 | break; |
a0d0e21e |
2677 | default: /* Called on something of 0 width. */ |
2678 | break; /* So match right here or not at all. */ |
2679 | } |
a687059c |
2680 | |
a0ed51b3 |
2681 | if (hardcount) |
2682 | c = hardcount; |
2683 | else |
2684 | c = scan - PL_reginput; |
3280af22 |
2685 | PL_reginput = scan; |
a687059c |
2686 | |
c277df42 |
2687 | DEBUG_r( |
2688 | { |
2689 | SV *prop = sv_newmortal(); |
2690 | |
2691 | regprop(prop, p); |
2692 | PerlIO_printf(Perl_debug_log, |
2693 | "%*s %s can match %ld times out of %ld...\n", |
2694 | REPORT_CODE_OFF+1, "", SvPVX(prop),c,max); |
2695 | }); |
2696 | |
a0d0e21e |
2697 | return(c); |
a687059c |
2698 | } |
2699 | |
2700 | /* |
c277df42 |
2701 | - regrepeat_hard - repeatedly match something, report total lenth and length |
2702 | * |
2703 | * The repeater is supposed to have constant length. |
2704 | */ |
2705 | |
76e3520e |
2706 | STATIC I32 |
cea2e8a9 |
2707 | S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) |
c277df42 |
2708 | { |
5c0ca799 |
2709 | dTHR; |
c277df42 |
2710 | register char *scan; |
2711 | register char *start; |
3280af22 |
2712 | register char *loceol = PL_regeol; |
a0ed51b3 |
2713 | I32 l = 0; |
708e3b05 |
2714 | I32 count = 0, res = 1; |
a0ed51b3 |
2715 | |
2716 | if (!max) |
2717 | return 0; |
c277df42 |
2718 | |
3280af22 |
2719 | start = PL_reginput; |
a0ed51b3 |
2720 | if (UTF) { |
708e3b05 |
2721 | while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { |
a0ed51b3 |
2722 | if (!count++) { |
2723 | l = 0; |
2724 | while (start < PL_reginput) { |
2725 | l++; |
2726 | start += UTF8SKIP(start); |
2727 | } |
2728 | *lp = l; |
2729 | if (l == 0) |
2730 | return max; |
2731 | } |
2732 | if (count == max) |
2733 | return count; |
2734 | } |
2735 | } |
2736 | else { |
708e3b05 |
2737 | while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { |
a0ed51b3 |
2738 | if (!count++) { |
2739 | *lp = l = PL_reginput - start; |
2740 | if (max != REG_INFTY && l*max < loceol - scan) |
2741 | loceol = scan + l*max; |
2742 | if (l == 0) |
2743 | return max; |
c277df42 |
2744 | } |
2745 | } |
2746 | } |
708e3b05 |
2747 | if (!res) |
3280af22 |
2748 | PL_reginput = scan; |
c277df42 |
2749 | |
a0ed51b3 |
2750 | return count; |
c277df42 |
2751 | } |
2752 | |
2753 | /* |
cb8d8820 |
2754 | - reginclass - determine if a character falls into a character class |
bbce6d69 |
2755 | */ |
2756 | |
76e3520e |
2757 | STATIC bool |
cea2e8a9 |
2758 | S_reginclass(pTHX_ register char *p, register I32 c) |
bbce6d69 |
2759 | { |
5c0ca799 |
2760 | dTHR; |
bbce6d69 |
2761 | char flags = *p; |
2762 | bool match = FALSE; |
2763 | |
2764 | c &= 0xFF; |
ae5c130c |
2765 | if (ANYOF_TEST(p, c)) |
bbce6d69 |
2766 | match = TRUE; |
2767 | else if (flags & ANYOF_FOLD) { |
2768 | I32 cf; |
2769 | if (flags & ANYOF_LOCALE) { |
3280af22 |
2770 | PL_reg_flags |= RF_tainted; |
22c35a8c |
2771 | cf = PL_fold_locale[c]; |
bbce6d69 |
2772 | } |
2773 | else |
22c35a8c |
2774 | cf = PL_fold[c]; |
ae5c130c |
2775 | if (ANYOF_TEST(p, cf)) |
bbce6d69 |
2776 | match = TRUE; |
2777 | } |
2778 | |
2779 | if (!match && (flags & ANYOF_ISA)) { |
3280af22 |
2780 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2781 | |
2782 | if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) || |
2783 | ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) || |
2784 | ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) || |
2785 | ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c))) |
2786 | { |
2787 | match = TRUE; |
2788 | } |
2789 | } |
2790 | |
ae5c130c |
2791 | return (flags & ANYOF_INVERT) ? !match : match; |
bbce6d69 |
2792 | } |
2793 | |
a0ed51b3 |
2794 | STATIC bool |
cea2e8a9 |
2795 | S_reginclassutf8(pTHX_ regnode *f, U8 *p) |
c485e607 |
2796 | { |
2797 | dTHR; |
a0ed51b3 |
2798 | char flags = ARG1(f); |
2799 | bool match = FALSE; |
2800 | SV *sv = (SV*)PL_regdata->data[ARG2(f)]; |
2801 | |
2802 | if (swash_fetch(sv, p)) |
2803 | match = TRUE; |
2804 | else if (flags & ANYOF_FOLD) { |
2805 | I32 cf; |
dfe13c55 |
2806 | U8 tmpbuf[10]; |
a0ed51b3 |
2807 | if (flags & ANYOF_LOCALE) { |
2808 | PL_reg_flags |= RF_tainted; |
2809 | uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); |
2810 | } |
2811 | else |
2812 | uv_to_utf8(tmpbuf, toLOWER_utf8(p)); |
2813 | if (swash_fetch(sv, tmpbuf)) |
2814 | match = TRUE; |
2815 | } |
2816 | |
2817 | if (!match && (flags & ANYOF_ISA)) { |
2818 | PL_reg_flags |= RF_tainted; |
2819 | |
2820 | if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) || |
2821 | ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) || |
2822 | ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) || |
2823 | ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p))) |
2824 | { |
2825 | match = TRUE; |
2826 | } |
2827 | } |
2828 | |
2829 | return (flags & ANYOF_INVERT) ? !match : match; |
2830 | } |
161b471a |
2831 | |
dfe13c55 |
2832 | STATIC U8 * |
cea2e8a9 |
2833 | S_reghop(pTHX_ U8 *s, I32 off) |
c485e607 |
2834 | { |
2835 | dTHR; |
a0ed51b3 |
2836 | if (off >= 0) { |
2837 | while (off-- && s < (U8*)PL_regeol) |
2838 | s += UTF8SKIP(s); |
2839 | } |
2840 | else { |
2841 | while (off++) { |
2842 | if (s > (U8*)PL_bostr) { |
2843 | s--; |
2844 | if (*s & 0x80) { |
2845 | while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) |
2846 | s--; |
2847 | } /* XXX could check well-formedness here */ |
2848 | } |
2849 | } |
2850 | } |
2851 | return s; |
2852 | } |
161b471a |
2853 | |
dfe13c55 |
2854 | STATIC U8 * |
cea2e8a9 |
2855 | S_reghopmaybe(pTHX_ U8* s, I32 off) |
a0ed51b3 |
2856 | { |
c485e607 |
2857 | dTHR; |
a0ed51b3 |
2858 | if (off >= 0) { |
2859 | while (off-- && s < (U8*)PL_regeol) |
2860 | s += UTF8SKIP(s); |
2861 | if (off >= 0) |
2862 | return 0; |
2863 | } |
2864 | else { |
2865 | while (off++) { |
2866 | if (s > (U8*)PL_bostr) { |
2867 | s--; |
2868 | if (*s & 0x80) { |
2869 | while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) |
2870 | s--; |
2871 | } /* XXX could check well-formedness here */ |
2872 | } |
2873 | else |
2874 | break; |
2875 | } |
2876 | if (off <= 0) |
2877 | return 0; |
2878 | } |
2879 | return s; |
2880 | } |